release

- term console via flatpak-spawn --host so host venvs resolve (bins.host_console_command)
- QSettings sync() before subprocess kill in choices/tested-refs dialogs
- console regression test: fails on the in-sandbox 0.2.1 console

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

DESIGN.md

@@ -114,11 +114,20 @@ To add a new API call usable from subprocesses:
 ### External interpreter resolution (`bins.py`)
 `src/testium/interpreter/utils/bins.py` — single source of truth for the paths to the external Python and Lua interpreters used by subprocesses.
 
-- `python_bin()` / `lua_bin()` : resolve once, cache in memory. User can override via the `python_bin` / `lua_bin` global dict keys (typically populated from the YAML config). Falls back to discovery on PATH (candidates: `python3`/`python` and `lua`/`lua5.5`/`lua5.4`/`lua5.3`/`lua5.2`/`lua5.1`).
+- `python_bin()` / `lua_bin()` : resolve and cache. The cache is keyed by `(name, override)` so that a later change to `gd[python_bin]` (typically when a `param.yaml` sets the key) triggers a re-resolution on the next lookup instead of returning the stale auto-discovered path. Falls back to discovery on PATH (candidates: `python3`/`python` and `lua`/`lua5.5`/`lua5.4`/`lua5.3`/`lua5.2`/`lua5.1`).
 - `ensure(*names)` : called by `TestSet._validate_runtime_deps()` at test load. Always requires `python` (the eval engine always runs); requires `lua` only if a `lua_func` item is in the tree. Fails fast with a clear error citing tried candidates and override key.
 
 Engines (`PyProcessBase`, `LuaProcessBase`, `EvalExecEngine`) call `bins.python_bin()`/`bins.lua_bin()` themselves — call sites never pass an explicit binary path.
 
+#### Override-timing contract (`apply_overrides`)
+`bins.python_bin()` is called for the **first** time inside `eval_process_init()` (the long-lived inline-`<| … |>` subprocess), which happens **before** the YAML param files are loaded. To make `-d python_bin=…` and the GUI `python_bin` preference take effect for `eval_proc` itself, `process.py:run()` applies them to gd **before** `eval_process_init()` via the `apply_overrides()` helper extracted from `update_global()`. The post-load `update_global()` call then re-applies the same overrides (after `prepare_global()` clears gd), keeping the gd value in sync with the cached resolution.
+
+| Override source | `eval_proc` | `py_func` / `cycle` / `post_exec` |
+|---|---|---|
+| `-d python_bin=…` (CLI) | ✅ | ✅ |
+| GUI `python_bin` preference | ✅ | ✅ |
+| `python_bin: …` in `param.yaml` | ❌ (eval_proc already started) | ✅ (cache re-resolves on key change) |
+
 ## Key files
 
 | Path | Role |
@@ -215,27 +224,66 @@ Four distribution channels coexist, all sharing the single `src/testium/` packag
 The `.deb` work-in-progress lives in `package/deb/`:
 - `test_distro.sh debian:bookworm | debian:trixie | ubuntu:24.04` spins up a Docker/Podman container, reports system package availability, falls back to pip for what's missing (`pyside6` on bookworm/ubuntu, `telnetlib3`, `junit_xml`), runs the validation suite. Currently green on the three targets.
 
+### Building all channels (`build_all.sh`)
+
+`build_all.sh` builds every artifact into `dist/` (manual PDF, wheel, PyInstaller binary, Flatpak bundle, AppImage). It reuses `scripts/build_env.sh` + `set_env.sh` so the venv at `test/tmp/.venv` stays the single source of Python deps; `build`/`pyinstaller`/`sphinx`/`linuxdoc` (and `pygls`, via the `[lsp]` extra) are installed there on demand. A step is skipped if its artifact already exists; `--clean` forces a rebuild.
+
+- **Parallelism (default).** A serial *prep* phase does everything that writes the shared venv (the `pip install`s) plus the Flatpak runtime install and the wheel (the AppImage installs it). Then manual + PyInstaller + Flatpak + AppImage build concurrently — they only *read* the venv, so there is no concurrent-pip race. Per-step output goes to `dist/.build-logs/<step>.log`; results print in completion order (`wait -n`), and a failing step's log is dumped at the end. `--serial` builds one at a time. Ctrl+C is trapped to kill each job's whole process tree (subshell + grandchildren: podman container, flatpak-builder, pyinstaller), so no orphans survive.
+- **`--ram` (slow/flash storage).** Redirects the build scratch to `/dev/shm` and skips UPX, a large win when building from a USB stick / SD card (I/O-bound on flash): `TMPDIR` + `PIP_CACHE_DIR`, the PyInstaller `--workpath` (`PYI_WORKPATH`), and a tmpfs bind-mount at the in-container AppImage AppDir (`APPIMAGE_APPDIR_TMPFS`); UPX is disabled via `TESTIUM_NO_UPX` (read by the `.spec`). **Flatpak is excluded** — `flatpak-builder` mounts its state dir with `rofiles-fuse` and FUSE cannot mount on `/dev/shm` (`fusermount: Permission denied`), so it builds on disk. Each `package/*/build.sh` honours these env vars with on-disk defaults, so behaviour is unchanged without `--ram`; the tmpfs scratch is freed on exit. On a RAM-limited machine combine with `--serial`.
+
 ### Host-only py_func / lua_func in sandboxed bundles (Flatpak, AppImage)
 
-The bundled Python (Flatpak's runtime python, AppImage's `python3.11`) is reserved for the **main process only**. Subprocesses (`py_func`, `lua_func`, `git`) must use the host's interpreters and tools so user-installed modules (pyserial, junit_xml, …) are visible. This is enforced by `interpreter/utils/bins.py`:
+The bundled Python (Flatpak's runtime python, AppImage's `python3.11`) is reserved for the **main process only**. Subprocesses (`py_func`, `lua_func`, `git`, the `run` item's sub-instance) must use the host's interpreters and tools so user-installed modules (pyserial, junit_xml, …) are visible. This is enforced by `interpreter/utils/bins.py`:
 
 - `_in_flatpak()` (checks `/.flatpak-info`) and `_in_appimage()` (checks `APPIMAGE` env var) detect the sandbox.
-- `_which(name)` probes only host bin dirs in those modes:
-  - Flatpak: `/run/host/usr/{local/,}bin`, `/run/host/bin` (host mounted via `--filesystem=host-os`).
-  - AppImage: `/usr/local/bin`, `/usr/bin`, `/bin` (we are directly on the host filesystem).
-  - If the host has no python3/lua, `ensure()` raises `ETUMRuntimeError` at test load with the candidate list — no silent fallback to a bundled interpreter.
-- User overrides (`python_bin`/`lua_bin` in globdict): bare names are resolved through `_which()` (host-only), absolute paths are accepted as-is.
-- `apply_host_libs(env)` is called by `py_process.py` / `lua_process.py` on the env passed to Popen:
-  - Flatpak: prepends host lib dirs to `LD_LIBRARY_PATH` so the dynamic linker finds host `.so`'s.
-  - AppImage: strips `$APPDIR`-prefixed entries from `LD_LIBRARY_PATH` / `PYTHONPATH` / `PATH` and drops `PYTHONHOME`, so the host Python doesn't try to load the bundled stdlib/site-packages.
-- `apply_host_lua_paths(env)` (Flatpak only) prepends `/run/host/usr/{lib,share}/lua/X.Y` to `LUA_PATH` / `LUA_CPATH` so `cjson`, `socket`, etc. resolve. Must be called **after** user `lua_env` overrides so host paths win. AppImage relies on host Lua's compiled-in defaults.
+- **Flatpak**: the sandbox glibc/ABI is incompatible with arbitrary host shared libraries, so we **cannot** run host binaries inside the Flatpak runtime — `LD_LIBRARY_PATH` injection trips a `_dl_call_libc_early_init` assertion. The supported way out is `flatpak-spawn --host`, a stub on `$PATH` inside every Flatpak that proxies an `exec` over D-Bus to the host's `org.freedesktop.Flatpak` service. The manifest grants `--talk-name=org.freedesktop.Flatpak` so the call is allowed. Helpers:
+  - `flatpak_host_spawn(interp, args, host_cwd, extra_env=…)` builds the spawn command vector with a curated set of forwarded env vars (`HOME`, `USER`, `DISPLAY`, `DBUS_SESSION_BUS_ADDRESS`, …) plus any explicit overrides.
+  - `_get_host_testium_path()` returns a path to the testium package the host can read. In Flatpak the package lives under `/app/lib/testium` which the host cannot see, so the package is staged once per process under `/tmp/testium_host_*` (`/tmp` is shared) and reused. In source / wheel / PyInstaller installs under `$HOME` the original path is returned untouched.
+  - `_which_host_flatpak(name)` resolves a binary by spawning `command -v` on the host (or `test -x` for absolute paths) — sandbox-visible probing under `/run/host/...` is unreliable (only `host-os` is mounted; user paths like `/scratch` aren't there).
+  - `_python_version()` and `_lua_version()` go through `_run_probe()` which dispatches to `flatpak-spawn` in Flatpak so validation happens against the actual host interpreter.
+  - `py_process.py` / `lua_process.py` `start()` use `flatpak_host_spawn` with `host_cwd = _get_host_testium_path()[+/lua_func]` and forward `PYTHONPATH` / `LUA_PATH` / `LUA_CPATH` / `PATH` as `--env=` arguments.
+  - The `run` item's `_testium_launch_cmd()` prefixes `flatpak run org.testium.Testium` with `flatpak-spawn --host` so the sub-instance is launched by the host's `flatpak` CLI, not by an unworkable in-sandbox `flatpak` binary.
+- **AppImage**: we are directly on the host filesystem, so the regular discovery on `/usr/local/bin`, `/usr/bin`, `/bin` suffices. `apply_host_libs(env)` strips `$APPDIR`-prefixed entries from `LD_LIBRARY_PATH` / `PYTHONPATH` / `PATH` and drops `PYTHONHOME` so the host Python doesn't try to load the bundled stdlib/site-packages.
+- User overrides (`python_bin`/`lua_bin` in globdict): in Flatpak, both bare names and absolute paths go through `_which()` so they are validated on the host side (the sandbox can't see e.g. `/scratch/...`). Outside Flatpak, absolute paths are accepted as-is and bare names go through PATH discovery.
+- If the host has no python3/lua, `ensure()` raises `ETUMRuntimeError` at test load with the candidate list — no silent fallback to a bundled interpreter.
 - `py_process.py` additionally pops `PYTHONUSERBASE` (set to `/var/data/python` by the Flatpak runtime, which would hide `~/.local/lib/...`).
 
+### Declarative test item parameters
+
+Each `TestItem` subclass declares its accepted parameters as a class attribute `PARAMS = ParamSet(Param(...), ...)` (`interpreter/utils/param_decl.py`). The descriptor carries the parameter name, *kind* (`SCALAR` — the default and may be omitted; `LIST`; `BLOCK`; `Enum("a", "b", ...)`), `required` flag, `default`, and free-form `doc`. There is **no Python type** in the descriptor on purpose: most parameter values are expressions (`$(...)` / `<| ... |>`) whose effective type is only known after expansion, so a static type would be misleading. Post-expansion `validate=lambda v: ...` callbacks are available as an opt-in for the rare cases where a runtime check is warranted (e.g. a specific format).
+
+`TestItem.COMMON_PARAMS` (in `test_item.py`) declares the 14 parameters accepted by every item: `name`, `doc`, `skipped`, `key`, `stop_on_failure`, `execute_on_stop`, `process_result`, `store_result`, `expected_result`, `no_fail`, `report`, `condition`, `steps`, and the internal `seq_filename` injected by the loader. The base class concatenates `COMMON_PARAMS + subclass.PARAMS` in `_validate_declared_params()` and:
+
+- emits a `tm.print_warn(...)` listing the accepted names when an unknown key appears in the user YAML (catches typos like `param_filee`);
+- raises `ETUMSyntaxError` (with the `.tum` source as context) when a `required=True` param is missing.
+
+Validation is **opt-in per subclass**: while a subclass keeps `PARAMS = None` (the base-class default), the check is skipped entirely. This kept the migration incremental — items can be visited one by one without forcing a big-bang change. All structured items have been migrated; only the "unstructured-body" classes (`TestItemConsoleWrite`/`WriteLn` which carry the message as the raw value, `TestItemPlotActionAdd`/`Export` which take arbitrary plot-data keys, `TestItemUnittestElement` which is internally instantiated with `dict_item=None`) intentionally remain unvalidated.
+
+Diagnostics are currently **warnings** for unknown params so an out-of-tree `.tum` with a pre-existing typo doesn't suddenly fail. The flip to a hard error is a one-line change in `_validate_declared_params()` once the user is comfortable.
+
+Action items follow the same declarative principle. A `TestItemActions` parent (`console`, `plot`, `json_rpc`) declares its nested actions as a class attribute `ACTIONS = {yaml_key: action_class}` (e.g. `{"open": TestItemConsoleOpen, "write": …}`), mirroring `PARAMS`. The base `TestItemActions.__init__` seeds `self.action_classes` from `type(self).ACTIONS`; the imperative `register_actions(**…)` method is retained only as an escape hatch for actions that can't be known at class-definition time (none today). Because the action classes are always defined above their parent in the module, the class-level dict resolves without forward-reference gymnastics.
+
+The schema is the realized source of truth for the LSP server (`testium lsp`), the `testium schema` CLI dump, and future auto-generated manual sections: `ParamSet.to_schema()` returns the JSON-Schema-shaped representation, and `lsp/schema.py` reads both `PARAMS` and `ACTIONS` **purely from class attributes** — no `inspect.getsource`/AST parsing. This is what lets the full schema (including nested actions) survive a frozen PyInstaller build where the `.py` source isn't on disk.
+
+### Language server (`testium lsp`) across channels
+
+The `testium_assist` editor extension is a thin LSP client that spawns `testium lsp` and talks JSON-RPC over stdio, so the language server must work from *every* distribution channel. Two requirements:
+
+1. **`pygls` (+ `lsprotocol`, `cattrs`, `attrs`, `typing_extensions`) must be bundled.** It is the pyproject `[lsp]` extra (kept optional so a plain `pip install testium` stays lean), wired into each full-app channel: `build_env.sh` installs it into the shared `test/tmp/.venv` (covers **source run** and the **PyInstaller** build env); the **AppImage** installs the wheel as `…whl[lsp]`; the **Flatpak** adds a `python3-lsp` pip module (network-at-build, consistent with the manifest's global `--share=network`); the **PyInstaller** `.spec` force-collects the submodules via `collect_submodules` + explicit `hiddenimports` (including the lazily-imported `lsp`, `lsp.server`, `lsp.schema`).
+2. **The schema must build without source** — handled by the declarative `PARAMS`/`ACTIONS` above; PyInstaller is the only channel that strips `.py` source, and it no longer matters.
+
+`test/validation/lsp_check.py` enforces both per channel: `run.sh` calls it before launching the suite, asserting that `<channel> schema` returns JSON whose `console`/`plot`/`json_rpc` items still carry their actions, and that `<channel> lsp` answers an `initialize` request with capabilities (and never reports the pygls dependency missing). So `./test/validation/run.sh --mode flatpak|pyinstaller|appimage` now fails loudly if a channel ships a broken or pygls-less language server.
+
 ### Version reporting (`interpreter/utils/version.py`)
 
 Both Flatpak and AppImage export `TESTIUM_VERSION` from a launcher (Flatpak: launcher script in `org.testium.Testium.yaml`; AppImage: `runtime.env` in `AppImageBuilder.yml`). `get_testium_version()` checks `/.flatpak-info` / `APPIMAGE` and reads `TESTIUM_VERSION` rather than relying on package metadata or repo introspection.
 
 ## Recent fixes / notable changes
+- `build_all.sh`: builds the four heavy channels in parallel (serial prep for the shared venv + wheel), results in completion order, Ctrl+C kills the whole job tree; `--ram` puts the build scratch on tmpfs (`/dev/shm`) + skips UPX for fast builds on USB/SD storage (Flatpak excluded — rofiles-fuse can't mount tmpfs). See the "Building all channels" section.
+- LSP across packaging channels: `testium lsp` (and the `testium_assist` editor extension that spawns it) now works from source, wheel, PyInstaller, Flatpak and AppImage. Two enablers — (1) action items declare a class-level `ACTIONS = {key: class}` registry (like `PARAMS`), so `lsp/schema.py` builds the full schema from class attributes with no `inspect.getsource`/AST (which broke under frozen PyInstaller); (2) the `[lsp]` extra (pygls) is wired into every full-app channel. `test/validation/lsp_check.py`, run by `run.sh` before the suite, asserts per-channel that `schema` keeps its actions and `lsp` answers `initialize`. See the matching architecture sections.
+- Declarative test item parameters (v0.2): each `TestItem` subclass exposes a `PARAMS = ParamSet(...)` class attribute consumed by the base `__init__`. Catches unknown YAML keys (typo warnings listing the accepted names) and missing required params (load-time errors with `.tum` context). Lays the schema foundation for a future LSP server and auto-generated manual sections. See the matching architecture section.
+- Flatpak: `py_func` / `lua_func` / `run` sub-instance now execute on the host via `flatpak-spawn --host`. The previous attempt to inject host lib dirs into the sandbox's `LD_LIBRARY_PATH` was abandoned — host shared libs are ABI-incompatible with the Flatpak runtime's glibc and would trip `_dl_call_libc_early_init`. The manifest gained `--talk-name=org.freedesktop.Flatpak` so the spawn proxy call is allowed. The testium package is staged once per process under `/tmp` (shared with the host) so the host interpreter can locate `py_func` / `lua_func`.
+- Validation suite: single entry point with `--mode source|wheel|pyinstaller|flatpak|appimage` to validate every packaging channel against the same items. Per-mode report filenames prevent clobbering.
 - Restructure: single `src/testium/` Python package (was 4 sibling top-levels: `testium`, `lib`, `py_func`, `lua_func`). `lib/` → `runtime/`, `libs/` → `api/`. `pip install` now produces a clean `site-packages/testium/` with no top-level pollution; `.lua` files travel via `package_data`.
 - `bins.py`: centralised resolution + cache of external `python3` / `lua` binaries. Replaces the scattered `tm.gd("python_bin")`/`tm.gd("lua_bin")` dance and the duplicated discovery logic in `py_process.py`/`lua_process.py`. Validates at test load via `TestSet._validate_runtime_deps()` so missing interpreters fail fast.
 - Subprocess API contract: user scripts in `py_func`/`lua_func` use the JSON-RPC bridge (`py_func.tm` / Lua `tm`) — never `api.testium` / `interpreter.*` directly. `SUPPORTED_API` extended with `OS`, `get_main_dir`, `init_timestamp`, `timestamp`, `timestamp_as_sec` so subprocess scripts have the same surface as main-process code.
@@ -261,12 +309,20 @@ Both Flatpak and AppImage export `TESTIUM_VERSION` from a launcher (Flatpak: lau
 - `unittest` item: renamed from `unittest_file`.
 - GUI test tree: check and fold state preserved across same-file reloads.
 - Licence: EUPL-1.2.
+- Interpreter override timing: `apply_overrides()` extracted from `update_global()` and called by `process.py:run()` before `eval_process_init()`, so `-d python_bin=…` / GUI prefs reach `bins.python_bin()` on its first lookup. `bins._resolve()` cache is now keyed by `(name, override)` so later `param.yaml` changes are picked up by subsequently constructed engines.
 
 ## Validation tests
-Located in `test/validation/`. Run with `-b` flag:
+Located in `test/validation/`. Two entry points:
 ```
-./run.sh -b -- test/validation/main.tum
+./test/validation/run.sh [clean] [--mode MODE] [extra args]   # wrapper — uses a dedicated venv (see below)
+./run.sh -b -- test/validation/main.tum                       # direct — testium's own python is used for test execution
 ```
+The same item set is reused across every packaging channel — `--mode source|wheel|pyinstaller|flatpak|appimage` selects which testium binary launches the suite (`source` is the default, invoking the project's `run.sh`). Each mode stamps its results into a distinct report file (`validation-<mode>.sqlite`, `validation-<mode>-<item>.xml`) so successive runs in different modes don't clobber each other. Prerequisites (PyInstaller binary built, Flatpak bundle installed, …) are checked before launch with a hint pointing at `build_all.sh`. On Windows only `source`, `wheel`, `pyinstaller` are supported.
+
+The `run.sh` / `run.bat` wrappers create a dedicated **host** Python venv at `${TMPDIR:-/tmp}/testium-validation-venv` (Linux) or `%TEMP%\testium-validation-venv` (Windows), with `--system-site-packages` + `pip install junit-xml`, and run the suite with `-d python_bin=…` so every test-execution subprocess (eval_proc, py_func, cycle, post_exec) runs inside that venv. testium itself keeps running in its own environment for the chosen mode. The venv is shared across modes because every test-execution subprocess ends up on the host either directly (source/wheel/pyinstaller/appimage) or via `flatpak-spawn --host` (flatpak). `clean` as the first argument recreates the venv. `wheel` mode also creates a separate `testium-wheel-venv-<v>` to hold the installed package.
+
+The `venv` item (`test/validation/items/venv/`) asserts that the override actually took effect: `python_bin` is set, `sys.executable` matches it, `sys.prefix == dirname(dirname(python_bin))`, and `sys.prefix != sys.base_prefix` (the last marker catches the case where `python_bin` happens to be a system interpreter, which path-equality alone would miss because the venv's `bin/python3` is a symlink to the host). Both `eval_proc` (inline `<| … |>`) and `py_func` paths are exercised.
+
 Parallel item tests: `test/validation/items/parallel/test.tum`
 
 ## Dependencies

README.md

@@ -27,6 +27,27 @@ Pre-built artifacts are published at
   runnable directly, no Python installation required on the host. Lua
   support still needs a system `lua` interpreter and the `lua-socket` /
   `lua-cjson` modules.
+* **AppImage** (`Testium-<version>-x86_64.AppImage`) — single-file
+  Linux binary, runnable directly:
+
+  ```sh
+  chmod +x Testium-*-x86_64.AppImage
+  ./Testium-*-x86_64.AppImage
+  ```
+
+  Requires `libfuse2` on the host (FUSE 2 — distinct from `fuse3`, which
+  most distros now ship by default):
+
+  | Distro | Package |
+  |--------|---------|
+  | Arch / CachyOS / Manjaro | `fuse2` |
+  | Debian trixie / Ubuntu 24.04+ | `libfuse2t64` |
+  | Debian bookworm / Ubuntu 22.04 | `libfuse2` |
+  | Fedora | `fuse-libs` |
+
+  If you can't install libfuse2 (e.g. minimal container), prefix the
+  invocation with `APPIMAGE_EXTRACT_AND_RUN=1` — the AppImage will
+  self-extract to `/tmp` on each run instead of FUSE-mounting.
 * **Flatpak bundle** (`testium.flatpak`) — install with:
 
   ```sh
@@ -41,6 +62,9 @@ Pre-built artifacts are published at
   `testium` command is available in the terminal (requires `~/.local/bin` in
   `PATH`, which most modern distributions provide by default).
 
+Every channel ships the language server, so `testium lsp` (see
+[Editor support](#editor-support)) works out of the box from any of them.
+
 ## Quick start
 
 From a checkout of the repository:
@@ -82,6 +106,45 @@ python3 src/testium               # GUI
 python3 src/testium -b mytest.tum # batch
 ```
 
+## Editor support
+
+testium ships a Language Server Protocol (LSP) server that gives `.tum` files
+completion of item types, hover documentation, and an outline view in any
+LSP-capable editor:
+
+```sh
+testium lsp        # speaks LSP over stdio; an editor's LSP client drives it
+testium schema     # dumps the item/parameter schema as JSON (what the LSP serves)
+```
+
+The server is bundled in every pre-built release (wheel, binary, Flatpak,
+AppImage). For a source / wheel install, pull the language-server extra:
+
+```sh
+pip install 'testium[lsp]'                 # from PyPI / a wheel
+pip install -e /path/to/testium/src[lsp]   # from a source checkout
+```
+
+A VSCode / VSCodium client extension (`testium_assist`) wraps `testium lsp`;
+the schema is built from testium itself, so new item types and parameters
+appear in the editor on the next testium upgrade with no client change.
+
+It is published on [Open VSX](https://open-vsx.org/extension/testium/testium-assist),
+so in **VSCodium, Cursor, Windsurf, Theia and code-server** it installs from the
+Extensions view (search `testium-assist`) or with
+`codium --install-extension testium.testium-assist`.
+
+**Microsoft VSCode** does not list Open VSX extensions, so install the `.vsix`
+by hand — download it from the Open VSX page above, then *Extensions → ⋯ →
+Install from VSIX…* or:
+
+```sh
+code --install-extension testium-assist-0.1.0.vsix
+```
+
+The extension runs `testium lsp`, so `testium` must be on the `PATH` (otherwise
+point the `testium.serverPath` setting at the binary/AppImage).
+
 ## Troubleshooting
 
 ### `wl_proxy_marshal_flags` symbol error

build_all.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
-# Build every distribution channel of testium, in order:
+# Build every distribution channel of testium:
 #   1. Manual PDF         -> dist/testium-manual-<v>.pdf
 #   2. Wheel              -> dist/testium-<v>-py3-none-any.whl   (PEP 427 name)
 #   3. PyInstaller binary -> dist/testium-<v>
@@ -8,6 +8,24 @@
 # release_note.txt is copied to dist/ up front (with a warning if it has no
 # entry for the current version).
 #
+# By default, a step is skipped if its artifact already exists in dist/.
+# Pass --clean to remove existing dist/ artifacts and rebuild everything.
+#
+# Parallelism: the wheel is built first (the AppImage installs it), then the
+# manual, PyInstaller, Flatpak and AppImage builds run concurrently. The shared
+# venv at test/tmp/.venv is only WRITTEN during the serial prep phase (the
+# `pip install` of build/sphinx/pyinstaller); the parallel builds only read it,
+# so there is no concurrent-pip race. Pass --serial to build one step at a time
+# (useful when debugging or on a resource-constrained machine). Per-step output
+# of the parallel phase is captured under dist/.build-logs/<step>.log and the
+# log of any failing step is printed at the end.
+#
+# Pass --ram to redirect the per-channel build scratch (PyInstaller workpath,
+# AppImage AppDir) and TMPDIR/PIP_CACHE_DIR to /dev/shm, and skip UPX. Big
+# speedup on slow/flash storage. Flatpak is excluded (its rofiles-fuse can't
+# mount on /dev/shm), so it still builds on disk. On a RAM-limited machine
+# combine with --serial (e.g. ./build_all.sh --ram --serial).
+#
 # All artifacts are collected (copied) under <repo>/dist/. Original outputs in
 # src/dist/, package/*/dist/, doc/manual/ are left in place. Wheel and AppImage
 # keep their original names (which already contain the version); manual,
@@ -15,17 +33,39 @@
 #
 # Re-uses scripts/build_env.sh and scripts/set_env.sh — the same pair invoked
 # by run.sh — so the venv at test/tmp/.venv stays the single source of Python
-# dependencies. `build` and `pyinstaller` are installed into that venv on
-# demand if not already there. Flatpak and AppImage build in their own
-# container/sandbox; their build.sh scripts have their own toolchain checks.
+# dependencies. `build`, `pyinstaller`, `sphinx` and `linuxdoc` are installed
+# into that venv on demand if not already there. Flatpak and AppImage build in
+# their own container/sandbox; their build.sh scripts have their own toolchain
+# checks.
 
 set -e
 
+CLEAN=0
+SERIAL=0
+RAM=0
+for arg in "$@"; do
+    case "$arg" in
+        --clean|-c) CLEAN=1 ;;
+        --serial)   SERIAL=1 ;;
+        --ram)      RAM=1 ;;
+        *) echo "Unknown option: $arg" >&2; exit 1 ;;
+    esac
+done
+
 SCRIPT_DIR=$(realpath "$(dirname "$0")")
 VERSION=$(cat "$SCRIPT_DIR/src/VERSION")
 DIST_DIR="$SCRIPT_DIR/dist"
 mkdir -p "$DIST_DIR"
 
+if [ "$CLEAN" -eq 1 ]; then
+    echo "-- clean: removing existing dist artifacts for version $VERSION"
+    rm -f "$DIST_DIR/testium-manual-${VERSION}.pdf"
+    rm -f "$DIST_DIR"/testium-${VERSION}-*.whl
+    rm -f "$DIST_DIR/testium-${VERSION}"
+    rm -f "$DIST_DIR/testium-${VERSION}.flatpak"
+    rm -f "$DIST_DIR"/Testium-${VERSION}-*.AppImage
+fi
+
 # Release note: copy it to dist/ and warn (but don't fail) if it has no entry
 # for the current version.
 RELEASE_NOTE_SRC="$SCRIPT_DIR/release_note.txt"
@@ -42,8 +82,30 @@ export REQ_PATH="$SCRIPT_DIR/src/requirements.txt"
 bash "$SCRIPT_DIR/scripts/build_env.sh"
 source "$SCRIPT_DIR/scripts/set_env.sh"
 
-# Ensure wheel/PyInstaller toolchains are present in the venv.
-python -m pip install --quiet --upgrade build pyinstaller
+# ---------- RAM mode: put build scratch on tmpfs (--ram) ----------------------
+# On slow storage (USB stick, SD card) the per-channel build dirs and temp
+# churn dominate. --ram redirects the PyInstaller workpath, the AppImage AppDir
+# and TMPDIR/PIP_CACHE_DIR to /dev/shm, and skips UPX. Flatpak is intentionally
+# NOT moved: flatpak-builder mounts its state dir with rofiles-fuse, and FUSE
+# can't mount on /dev/shm (fusermount: Permission denied) — so it builds on
+# disk. The tmpfs scratch is freed on exit.
+if [ "$RAM" -eq 1 ]; then
+    RAMROOT="/dev/shm/testium-build-${VERSION}"
+    echo "-- RAM mode: build scratch under $RAMROOT (tmpfs), freed on exit"
+    echo "   (flatpak builds on disk — rofiles-fuse can't mount on /dev/shm)"
+    rm -rf "$RAMROOT"
+    mkdir -p "$RAMROOT"/{tmp,pip,pyi-work,appdir}
+    export TMPDIR="$RAMROOT/tmp"
+    export PIP_CACHE_DIR="$RAMROOT/pip"
+    export PYI_WORKPATH="$RAMROOT/pyi-work"            # pyinstaller --workpath
+    export APPIMAGE_APPDIR_TMPFS="$RAMROOT/appdir"     # AppDir bind-mount
+    export TESTIUM_NO_UPX=1                            # skip slow UPX in the spec
+    trap 'rm -rf "$RAMROOT"' EXIT
+    if [ "$SERIAL" -ne 1 ]; then
+        echo "   note: with --ram, prefer adding --serial so each step gets the"
+        echo "         full tmpfs and you don't risk OOM (flatpak+appimage are ~1 GB each)."
+    fi
+fi
 
 step() {
     echo
@@ -52,56 +114,186 @@ step() {
     echo "================================================================"
 }
 
-# 1. Manual PDF
-step "1/5  Manual PDF (version $VERSION)"
-bash "$SCRIPT_DIR/doc/manual/sphinx/build_doc.sh"
-MANUAL_SRC="$SCRIPT_DIR/doc/manual/testium_manual.pdf"
+# Kill a process and its whole descendant tree (children first) — used by the
+# interrupt handler so SIGINT also stops grandchildren the parallel jobs spawned
+# (podman container, flatpak-builder, pyinstaller …), not just the subshells.
+_kill_tree() {
+    local pid=$1 c
+    for c in $(pgrep -P "$pid" 2>/dev/null); do
+        _kill_tree "$c"
+    done
+    kill -TERM "$pid" 2>/dev/null || true
+}
+
+# Set as INT/TERM handler around the parallel wait. Stops every running build
+# tree, then exits — the EXIT trap (set under --ram) frees the tmpfs scratch.
+_interrupt() {
+    echo >&2
+    echo "-- interrupted: stopping running builds…" >&2
+    local pid
+    for pid in "${!PID2NAME[@]}"; do
+        _kill_tree "$pid"
+    done
+    exit 130
+}
+
+# ---------- artifact paths ----------------------------------------------------
+
 MANUAL="$DIST_DIR/testium-manual-${VERSION}.pdf"
-cp -f "$MANUAL_SRC" "$MANUAL"
-
-# 2. Wheel — PEP 427 name kept (already contains version)
-step "2/5  Wheel (version $VERSION)"
-(
-    cd "$SCRIPT_DIR/src"
-    rm -rf dist build *.egg-info
-    python -m build --wheel
-)
-WHEEL_SRC=$(ls -1t "$SCRIPT_DIR/src/dist"/*.whl | head -1)
-WHEEL="$DIST_DIR/$(basename "$WHEEL_SRC")"
-cp -f "$WHEEL_SRC" "$WHEEL"
-
-# 3. PyInstaller binary
-step "3/5  PyInstaller binary (version $VERSION)"
-bash "$SCRIPT_DIR/package/pyinstaller/build.sh"
-PYI_SRC="$SCRIPT_DIR/package/pyinstaller/dist/testium"
 PYI_BIN="$DIST_DIR/testium-${VERSION}"
-cp -f "$PYI_SRC" "$PYI_BIN"
-
-# 4. Flatpak bundle
-step "4/5  Flatpak bundle (version $VERSION)"
-(
-    cd "$SCRIPT_DIR/package/flatpak"
-    bash build.sh
-)
-FLATPAK_SRC="$SCRIPT_DIR/package/flatpak/testium.flatpak"
 FLATPAK_BUNDLE="$DIST_DIR/testium-${VERSION}.flatpak"
-cp -f "$FLATPAK_SRC" "$FLATPAK_BUNDLE"
+wheel_in_dist() { ls -1t "$DIST_DIR"/testium-${VERSION}-*.whl 2>/dev/null | head -1; }
+appimage_in_dist() { ls -1t "$DIST_DIR"/Testium-${VERSION}-*.AppImage 2>/dev/null | head -1; }
 
-# 5. AppImage
-step "5/5  AppImage (version $VERSION)"
-(
-    cd "$SCRIPT_DIR/package/appimage"
-    bash build.sh
-)
-APPIMAGE_SRC=$(ls -1t "$SCRIPT_DIR/package/appimage"/*.AppImage 2>/dev/null | head -1)
-APPIMAGE="$DIST_DIR/$(basename "$APPIMAGE_SRC")"
-cp -f "$APPIMAGE_SRC" "$APPIMAGE"
-chmod +x "$APPIMAGE"
+# ---------- per-step build functions (assume tools are installed) -------------
+
+build_wheel() {
+    if [ -n "$(wheel_in_dist)" ]; then echo "wheel: already built — skipping"; return 0; fi
+    echo "wheel: building"
+    (
+        cd "$SCRIPT_DIR/src"
+        rm -rf dist build *.egg-info
+        python -m build --wheel
+    )
+    local src; src=$(ls -1t "$SCRIPT_DIR/src/dist"/*.whl | head -1)
+    cp -f "$src" "$DIST_DIR/$(basename "$src")"
+    echo "wheel: done"
+}
+
+build_manual() {
+    if [ -f "$MANUAL" ]; then echo "manual: already built — skipping"; return 0; fi
+    echo "manual: building"
+    bash "$SCRIPT_DIR/doc/manual/sphinx/build_doc.sh"
+    cp -f "$SCRIPT_DIR/doc/manual/testium_manual.pdf" "$MANUAL"
+    echo "manual: done"
+}
+
+build_pyinstaller() {
+    if [ -f "$PYI_BIN" ]; then echo "pyinstaller: already built — skipping"; return 0; fi
+    echo "pyinstaller: building"
+    bash "$SCRIPT_DIR/package/pyinstaller/build.sh"
+    cp -f "$SCRIPT_DIR/package/pyinstaller/dist/testium" "$PYI_BIN"
+    echo "pyinstaller: done"
+}
+
+build_flatpak() {
+    if [ -f "$FLATPAK_BUNDLE" ]; then echo "flatpak: already built — skipping"; return 0; fi
+    echo "flatpak: building"
+    (
+        cd "$SCRIPT_DIR/package/flatpak"
+        bash build.sh
+    )
+    cp -f "$SCRIPT_DIR/package/flatpak/testium.flatpak" "$FLATPAK_BUNDLE"
+    echo "flatpak: done"
+}
+
+build_appimage() {
+    if [ -n "$(appimage_in_dist)" ]; then echo "appimage: already built — skipping"; return 0; fi
+    echo "appimage: building"
+    (
+        cd "$SCRIPT_DIR/package/appimage"
+        bash build.sh
+    )
+    local src; src=$(ls -1t "$SCRIPT_DIR/package/appimage"/*.AppImage 2>/dev/null | head -1)
+    cp -f "$src" "$DIST_DIR/$(basename "$src")"
+    chmod +x "$DIST_DIR/$(basename "$src")"
+    echo "appimage: done"
+}
+
+# ---------- serial prep: tool installs (shared venv) + flatpak runtimes -------
+
+step "Prep: build tools + runtimes (serial — shared venv)"
+
+[ -f "$MANUAL" ]   || python -m pip install --quiet --upgrade sphinx linuxdoc
+[ -n "$(wheel_in_dist)" ] || python -m pip install --quiet --upgrade build
+[ -f "$PYI_BIN" ]  || python -m pip install --quiet --upgrade pyinstaller
+
+if [ ! -f "$FLATPAK_BUNDLE" ]; then
+    FLATPAK_DEPS=(
+        "org.kde.Platform//6.10"
+        "org.kde.Sdk//6.10"
+        "io.qt.PySide.BaseApp//6.10"
+    )
+    if ! flatpak remotes --user | grep -q "^flathub"; then
+        echo "  Adding Flathub remote"
+        flatpak remote-add --user --if-not-exists flathub https://dl.flathub.org/repo/flathub.flatpakrepo
+    fi
+    for dep in "${FLATPAK_DEPS[@]}"; do
+        if ! flatpak info --user "$dep" &>/dev/null && ! flatpak info --system "$dep" &>/dev/null; then
+            echo "  Installing Flatpak dependency: $dep"
+            flatpak install --user --noninteractive flathub "$dep"
+        fi
+    done
+fi
+
+# ---------- serial: wheel (the AppImage installs it) --------------------------
+
+step "1/5  Wheel (version $VERSION)"
+build_wheel
+
+# ---------- build the rest --------------------------------------------------
+
+REST=(manual pyinstaller flatpak appimage)
+
+if [ "$SERIAL" -eq 1 ]; then
+    n=2
+    for name in "${REST[@]}"; do
+        step "$n/5  $name (version $VERSION)"
+        "build_$name"
+        n=$((n + 1))
+    done
+else
+    step "2-5/5  manual + pyinstaller + flatpak + appimage (parallel)"
+    LOGDIR="$DIST_DIR/.build-logs"
+    mkdir -p "$LOGDIR"
+    declare -A PID2NAME
+    for name in "${REST[@]}"; do
+        log="$LOGDIR/$name.log"
+        echo "  -> launching $name (log: $log)"
+        ( "build_$name" ) >"$log" 2>&1 &
+        PID2NAME[$!]="$name"
+    done
+
+    # From here until all jobs are reaped, Ctrl+C stops every build tree.
+    trap _interrupt INT TERM
+
+    # Reap in completion order (wait -n) so each result prints the moment that
+    # build finishes, not when its slot comes up in the array.
+    FAILED=()
+    remaining=${#PID2NAME[@]}
+    while [ "$remaining" -gt 0 ]; do
+        if wait -n -p donepid; then rc=0; else rc=$?; fi
+        name="${PID2NAME[$donepid]:-}"
+        [ -z "$name" ] && continue
+        if [ "$rc" -eq 0 ]; then
+            echo "  -> $name: OK"
+        else
+            echo "  -> $name: FAILED (rc=$rc)"
+            FAILED+=("$name")
+        fi
+        remaining=$((remaining - 1))
+    done
+
+    trap - INT TERM
+
+    if [ "${#FAILED[@]}" -gt 0 ]; then
+        for name in "${FAILED[@]}"; do
+            echo
+            echo "===================== $name log ====================="
+            cat "$LOGDIR/$name.log"
+        done
+        echo >&2
+        echo "BUILD FAILED: ${FAILED[*]} (logs under $LOGDIR)" >&2
+        exit 1
+    fi
+fi
+
+# ---------- summary -----------------------------------------------------------
 
 step "All packages built"
 printf "  manual       : %s\n" "$MANUAL"
-printf "  wheel        : %s\n" "$WHEEL"
+printf "  wheel        : %s\n" "$(wheel_in_dist)"
 printf "  pyinstaller  : %s\n" "$PYI_BIN"
 printf "  flatpak      : %s\n" "$FLATPAK_BUNDLE"
-printf "  appimage     : %s\n" "$APPIMAGE"
+printf "  appimage     : %s\n" "$(appimage_in_dist)"
 printf "  release_note : %s\n" "$RELEASE_NOTE"

doc/manual/sphinx/source/modes.rst

@@ -23,3 +23,80 @@ graphical interface.
     :caption: call a test in batch mode
 
     testium -b test/my_test/main.tum
+
+.. _sec_language_server:
+
+Language server (editor support)
+--------------------------------
+
+*testium* ships a `Language Server Protocol
+<https://microsoft.github.io/language-server-protocol/>`_ server so that
+``.tum`` files get editor assistance — completion of test item types, hover
+documentation of their parameters, and an outline view — in any LSP-capable
+editor.
+
+The server speaks LSP over standard input/output and is started with:
+
+.. code-block:: text
+    :caption: start the language server
+
+    testium lsp
+
+It is not meant to be launched directly by the user: an editor's LSP client
+spawns it and drives the exchange. A VSCode / VSCodium client extension,
+*testium_assist*, is provided for that purpose; any other LSP client (Neovim,
+Emacs ``lsp-mode``, …) can be pointed at ``testium lsp`` as well.
+
+The information the server exposes is the test item schema, which can also be
+dumped as JSON for inspection or tooling:
+
+.. code-block:: text
+    :caption: dump the item / parameter schema
+
+    testium schema
+
+Because the schema is built from *testium* itself, every new item type or
+parameter becomes available in the editor on the next *testium* upgrade, with
+no change to the client.
+
+The language server is included in the pre-built binary, Flatpak and AppImage
+releases. For a source or wheel installation, pull the optional ``lsp``
+dependencies:
+
+.. code-block:: text
+    :caption: enable the language server for a wheel / source install
+
+    pip install 'testium[lsp]'
+
+Installing the VSCode / VSCodium extension
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+The *testium_assist* client extension is published on `Open VSX
+<https://open-vsx.org/extension/testium/testium-assist>`_, the registry used by
+VSCodium, Cursor, Windsurf, Eclipse Theia and code-server. In those editors,
+open the Extensions view and search ``testium-assist``, or install it from the
+command line:
+
+.. code-block:: text
+    :caption: install in VSCodium and other Open VSX editors
+
+    codium --install-extension testium.testium-assist
+
+Microsoft *VSCode* uses a different marketplace that does not list Open VSX
+extensions, so install the packaged ``.vsix`` by hand. Download it from the
+Open VSX page linked above, then either choose *Extensions* → *⋯* →
+*Install from VSIX…* in the UI, or run:
+
+.. code-block:: text
+    :caption: install the .vsix in Microsoft VSCode
+
+    code --install-extension testium-assist-0.1.0.vsix
+
+The extension launches ``testium lsp``, so the ``testium`` command must be on
+the ``PATH``. If *testium* is installed elsewhere — a specific binary or an
+AppImage — point the ``testium.serverPath`` setting at it instead.
+
+Once installed, open a ``.tum`` file: completion of item types, hover
+documentation and the outline view become available. If nothing happens, check
+that no ``files.associations`` entry forces ``*.tum`` to another language (it
+must stay the ``tum`` language the extension provides).

doc/manual/sphinx/source/tum_syntax.rst

@@ -232,6 +232,15 @@ list of the main test item (and eventually of the loop test item).
 TUM file ``main`` item is itself a variant of test items with a name and an
 step list attributes.
 
+.. note::
+
+   Each test item declares the parameters it accepts. When a ``.tum`` file
+   uses a key the item does not know, *testium* emits a warning listing the
+   accepted parameter names (catching typos such as ``param_filee`` for
+   ``param_file``); a missing **required** parameter aborts loading with an
+   error pointing at the source ``.tum`` file. Valid existing tests are
+   unaffected.
+
 
 
 .. toctree::

package/appimage/AppImageBuilder.yml

@@ -77,7 +77,10 @@ AppDir:
         python3.11 -m pip install --break-system-packages --upgrade --isolated --no-input --ignore-installed --prefix=$TARGET_APPDIR/usr -r ../../src/requirements.txt
 
         export PIP_CONFIG_FILE=$HOME/.pip/pip.conf
-        python3.11 -m pip install --break-system-packages --upgrade --isolated --no-input --ignore-installed --prefix=$TARGET_APPDIR/usr ../../src/dist/testium-{{APP_VERSION}}-py3-none-any.whl
+        # Install the wheel with the [lsp] extra so `testium lsp` (pygls) works
+        # from the AppImage. The extra pulls pygls/lsprotocol/cattrs/attrs from
+        # the index (network is available at build time, see get-pip above).
+        python3.11 -m pip install --break-system-packages --upgrade --isolated --no-input --ignore-installed --prefix=$TARGET_APPDIR/usr "../../src/dist/testium-{{APP_VERSION}}-py3-none-any.whl[lsp]"
 
 
 AppImage:

package/appimage/build.sh

@@ -17,11 +17,20 @@ else
 fi
 echo "Using $RUNTIME — building testium $APP_VERSION AppImage..."
 
+# APPIMAGE_APPDIR_TMPFS (set by build_all --ram) bind-mounts a host tmpfs dir at
+# the AppDir build path, keeping the ~1 GB AppDir churn off slow storage.
+APPDIR_MOUNT=""
+if [ -n "$APPIMAGE_APPDIR_TMPFS" ]; then
+    mkdir -p "$APPIMAGE_APPDIR_TMPFS"
+    APPDIR_MOUNT="-v $APPIMAGE_APPDIR_TMPFS:/work/package/appimage/AppDir"
+fi
+
 # APPIMAGE_EXTRACT_AND_RUN=1 lets appimagetool run without FUSE in the container.
 $RUNTIME run --rm \
     --privileged \
     -e APPIMAGE_EXTRACT_AND_RUN=1 \
     -v "$REPO_ROOT:/work" \
+    $APPDIR_MOUNT \
     -w /work/package/appimage \
     debian:bookworm bash -c "
         set -e

package/flatpak/build.sh

@@ -7,11 +7,19 @@
 
 set -e
 
-# Build + install local
-flatpak-builder --user --verbose --force-clean --install --repo=repo build org.testium.Testium.yaml
+# Build + install local. FLATPAK_BUILDDIR / FLATPAK_STATEDIR / FLATPAK_REPODIR
+# (set by build_all --ram) redirect the build dir, the state dir
+# (.flatpak-builder) and the ostree repo to tmpfs. flatpak-builder hardlinks
+# between the state dir and the build dir, so they MUST be on the same
+# filesystem — hence the state dir moves to tmpfs too (its download cache then
+# doesn't persist across --ram runs).
+BUILDDIR="${FLATPAK_BUILDDIR:-build}"
+STATEDIR="${FLATPAK_STATEDIR:-.flatpak-builder}"
+REPODIR="${FLATPAK_REPODIR:-repo}"
+flatpak-builder --user --verbose --force-clean --install --state-dir="$STATEDIR" --repo="$REPODIR" "$BUILDDIR" org.testium.Testium.yaml
 
 # Génère le bundle distribuable
-flatpak build-bundle repo testium.flatpak org.testium.Testium
+flatpak build-bundle "$REPODIR" testium.flatpak org.testium.Testium
 echo "Bundle généré : $(pwd)/testium.flatpak"
 
 # Crée ~/.local/bin/testium pour pouvoir taper "testium" en console

package/flatpak/org.testium.Testium.yaml

@@ -16,6 +16,11 @@ finish-args:
   - --filesystem=home
   - --filesystem=/tmp
   - --filesystem=host-os
+  # Allow flatpak-spawn --host to launch host binaries (Python, Lua, git…)
+  # outside the sandbox. Required because the sandbox glibc/ABI is
+  # incompatible with arbitrary host shared libraries — we route py_func and
+  # lua_func through the host instead.
+  - --talk-name=org.freedesktop.Flatpak
 
 build-options:
   build-args:
@@ -23,6 +28,23 @@ build-options:
 
 modules:
   - python3-requirements.json
+
+  # Language-server deps for `testium lsp` (pygls + lsprotocol + cattrs + attrs
+  # + typing_extensions). Installed from PyPI at build time — the build already
+  # runs with --share=network (see build-options). The core runtime deps stay
+  # offline-pinned in python3-requirements.json; these are pure-python wheels,
+  # hence --only-binary=:all: (no compilation, deterministic).
+  - name: python3-lsp
+    buildsystem: simple
+    build-options:
+      build-args:
+        - --share=network
+    build-commands:
+      # Whole command single-quoted: the ':all: ' colon-space would otherwise
+      # make YAML parse this list item as a mapping, silently dropping the
+      # command (flatpak-builder then runs an empty module — installs nothing).
+      - 'pip3 install --prefix=${FLATPAK_DEST} --only-binary=:all: "pygls>=1.3"'
+
   # 1. Dépendances Python tierces (HORS PySide6)
   # Utilisez flatpak-pip-generator pour vos autres libs (ex: pyserial, requests, etc.)
   # - name: python3-requirements

package/pyinstaller/build.sh

@@ -2,11 +2,15 @@
 
 SCRIPT_DIR=$(realpath $( dirname "$0"))
 
-rm -r "${SCRIPT_DIR}/build" "${SCRIPT_DIR}/dist"
+rm -rf "${SCRIPT_DIR}/build" "${SCRIPT_DIR}/dist"
 
 pwd=$(pwd)
 cd ${SCRIPT_DIR}
-pyinstaller testium.spec
+# PYI_WORKPATH (set by build_all --ram) puts the big intermediate build tree on
+# tmpfs; dist/ stays local so build_all can collect the binary.
+WORKARG=""
+[ -n "$PYI_WORKPATH" ] && WORKARG="--workpath $PYI_WORKPATH"
+pyinstaller $WORKARG testium.spec
 RESULT=$?
 if [ -n "$1" ] && [ "$1" = "install" ]; then
     if [ $RESULT -eq 0 ]; then

package/pyinstaller/testium.spec

@@ -1,5 +1,21 @@
 # -*- mode: python ; coding: utf-8 -*-
 import os
+from PyInstaller.utils.hooks import collect_submodules
+
+# Language-server dependencies for `testium lsp`. pygls/lsprotocol register
+# converters and features dynamically, so we collect their submodules wholesale
+# and force-import their pure-python deps (cattrs/attrs/typing_extensions).
+# The testium lsp modules are imported lazily by the CLI dispatch
+# (`from lsp.server import serve`), which PyInstaller's static analysis misses —
+# hence the explicit names. No source files need bundling: the schema export is
+# now fully declarative (PARAMS + ACTIONS class attributes), so it no longer
+# reads .py source via inspect.getsource (which fails in a frozen build).
+_LSP_HIDDEN = (
+    collect_submodules("pygls")
+    + collect_submodules("lsprotocol")
+    + ["cattrs", "attr", "attrs", "typing_extensions",
+       "lsp", "lsp.server", "lsp.schema"]
+)
 
 # junit_xml is imported by post_exec scripts running under the *host* Python,
 # not the frozen interpreter — so bundling it via hiddenimports alone is not
@@ -54,7 +70,7 @@ a = Analysis(
                    "colorama",
                    "matplotlib",
                    "junit_xml",
-                   "lxml"],
+                   "lxml"] + _LSP_HIDDEN,
     hookspath=[],
     hooksconfig={},
     runtime_hooks=[],
@@ -73,7 +89,9 @@ exe = EXE(
     debug=False,
     bootloader_ignore_signals=False,
     strip=False,
-    upx=True,
+    # UPX is CPU+IO heavy for a marginal size gain — build_all --ram sets
+    # TESTIUM_NO_UPX=1 to skip it (much faster on slow/flash storage).
+    upx=not os.environ.get("TESTIUM_NO_UPX"),
     upx_exclude=[],
     runtime_tmpdir=None,
     console=True,

release_note.txt

@@ -1,3 +1,68 @@
+version 0.2.2
+==============
+- Flatpak sandbox issue fixed for term console. Now a term console is
+  exactly like a host console.
+- Persistence fix of dialogs in case of flatpak.
+
+version 0.2.1
+==============
+- Faster test loading, especially for large tests built from jinja
+  templates and ``!include``: compiled jinja templates are cached and
+  reused (a file included many times is compiled once), rendering happens
+  in memory instead of through a temporary file, and YAML is parsed with
+  the libyaml C loader when available. Typical load time is 3-6x lower on
+  include / template-heavy tests; behaviour is unchanged.
+- Fix: a nested list holding more than one step under ``steps`` no longer
+  duplicates its entries while the step tree is built.
+- New load-time benchmark under ``test/benchmark/`` (synthetic-tree
+  generator + in-process timing harness) to measure the load pipeline.
+
+version 0.2
+==============
+- Test items: each item type now declares its accepted parameters
+  (``PARAMS = ParamSet(...)``). Typos in a ``.tum`` are surfaced as a
+  WARN listing the accepted names instead of being silently ignored;
+  missing required parameters error out at load time with the source
+  ``.tum`` file as context. No change to valid existing tests.
+- Editor support: testium now ships a language server. ``testium lsp``
+  gives ``.tum`` files item-type completion, hover documentation and an
+  outline view in any LSP-capable editor (a VSCode / VSCodium client is
+  provided separately). ``testium schema`` dumps the item/parameter
+  schema as JSON. The server works from every channel — bundled in the
+  binary / Flatpak / AppImage, and pulled by ``pip install testium[lsp]``
+  for wheel installs.
+- build_all.sh: the four heavy channels now build in parallel (results
+  reported as each finishes; Ctrl+C stops them cleanly). New ``--ram``
+  option builds on a tmpfs (``/dev/shm``) and skips UPX for much faster
+  packaging on USB-stick / SD-card storage.
+
+version 0.1.3
+==============
+- Stop interrupts engaged blocking steps (console, py_func, lua_func,
+  json_rpc, sleep) within ~200 ms instead of waiting for the step
+  to finish.
+- GUI Start / Stop / Pause flow simplified.
+- lua_func: a function returning nil is no longer reported as a failure.
+- ``-d python_bin=...`` and the GUI ``python_bin`` preference now reach
+  the eval subprocess (used to be silently ignored). ``param.yaml`` can
+  also override ``python_bin`` for py_func / cycle / post_exec.
+- Validation suite: ``test/validation/run.sh`` (and ``run.bat``)
+  runs the suite inside a dedicated venv in the system temp dir.
+- build_all.sh: ``release_note.txt`` and the user manual copied into
+  ``dist/``; warning if the file has no entry for the version being built.
+- Flatpak: every GUI file/directory dialog (open test, save report, log
+  path, default report/log dirs, python/lua interpreter pickers) now
+  bypasses the XDG document portal — the v0.1.2 fix was only on the
+  "open test" dialog.
+- Flatpak: py_func / lua_func / run sub-instance now execute on the host
+  via flatpak-spawn, lifting the previous glibc/ABI incompatibility that
+  prevented user-configured host Python or Lua interpreters from being
+  reached from the sandbox.
+- Validation suite: single entry point with ``--mode source|wheel|
+  pyinstaller|flatpak|appimage`` to validate any packaging channel
+  against the same item set; reports are stamped per mode.
+- GUI: the "Run tum" test item now uses the testium logo.
+
 version 0.1.2
 ==============
 - Flatpak: opening a test from the GUI now correctly finds its companion

scripts/build_env.sh

@@ -33,6 +33,11 @@ if [ ! -d "$PY_VENV_DIR" ]; then
     python3 -m venv "$PY_VENV_DIR"
     source "$PY_VENV_DIR/bin/activate"
     pip install --extra-index-url https://pypi.python.org/pypi -r $REQ_PATH
+    # Language-server deps (the pyproject [lsp] extra). Installed here so the
+    # source run AND the PyInstaller build — both of which use this venv — can
+    # start / collect the `testium lsp` server. pip-installed wheel users get
+    # them via `pip install testium[lsp]` instead.
+    pip install --extra-index-url https://pypi.python.org/pypi "pygls>=1.3"
     # Validation suite plugin used to verify the report-exporter
     # entry-points discovery end-to-end.
     FAKE_EXPORTER_DIR="$(dirname "$REQ_PATH")/../test/validation/fake_exporter"

src/VERSION

@@ -1 +1 @@
-0.1.2
+0.2.2

src/pyproject.toml

@@ -30,6 +30,12 @@ dependencies = [
 ]
 dynamic = ["version"]
 
+[project.optional-dependencies]
+# `pip install testium[lsp]` adds the language-server dependencies. The
+# stdio-only LSP server (`testium lsp`) reuses the schema export from the
+# core install; pygls is the only marginal cost.
+lsp = ["pygls>=1.3"]
+
 [project.scripts]
 testium = "testium:main"
 

src/testium/__init__.py

@@ -11,6 +11,30 @@ sys.path.append(os.path.abspath(ourpath.parent))
 import interpreter.utils.constants as cst
 
 def main():
+    # Subcommand dispatch (must run *before* argparse so neither 'schema' nor
+    # 'lsp' has to share the GUI/batch flag surface). The subcommands also
+    # skip the multiprocessing 'spawn' setup which is only meaningful for the
+    # main runtime — schema is a pure stdout dump and lsp speaks JSON-RPC
+    # over stdio without ever forking a test process.
+    if len(sys.argv) >= 2 and sys.argv[1] in ("schema", "lsp"):
+        sub = sys.argv[1]
+        if sub == "schema":
+            from lsp.schema import dump_all_schemas_json
+            print(dump_all_schemas_json())
+            return
+        # lsp
+        try:
+            from lsp.server import serve
+        except ImportError as e:
+            print(
+                f"testium lsp: language server dependencies missing ({e.name}). "
+                "Install with: pip install 'testium[lsp]'",
+                file=sys.stderr,
+            )
+            sys.exit(2)
+        serve()
+        return
+
     # This line sets the method for the "Process" function. It is required for Linux
     # support of the test dialogs.
     multiprocessing.set_start_method('spawn')

src/testium/api/termconsole.py

@@ -81,9 +81,13 @@ class TermConsole(Console):
                                    bufsize=0)
 
         else:
-            self.term = pexpect.spawn(  shell_cmd,
-                                        echo=False,
-                                        cwd=self.ppath)
+            # In Flatpak this returns a `flatpak-spawn --host` wrapper so the
+            # console behaves like a host shell (matching py_func / lua_func /
+            # run); elsewhere it's the chosen command unchanged.
+            from interpreter.utils import bins
+            argv = bins.host_console_command(shell_cmd, self.ppath)
+            self.term = pexpect.spawn(argv[0], args=argv[1:],
+                                      echo=False, cwd=self.ppath)
 
         self.q = BytesStore()
         self.t = threading.Thread(target=self.enqueue_output)

src/testium/interpreter/process.py

@@ -16,6 +16,7 @@ from interpreter.utils.test_init import (
     env_init,
     prepare_global,
     update_global,
+    apply_overrides,
     set_standard_gd_keys,
     test_run_init,
     test_run_header,
@@ -210,6 +211,19 @@ class TestProcess(Process):
 
                 env_init()
 
+                # Apply GUI defaults and CLI defines to the global dict
+                # *before* eval_proc starts: bins.python_bin() reads
+                # ``python_bin`` from gd on its very first call (during
+                # eval_process_init) and caches the result. Without this,
+                # ``-d python_bin=...`` and the GUI ``python_bin`` preference
+                # would only take effect for items spawned *after* the cache
+                # was already populated with the auto-discovered interpreter,
+                # i.e. they would silently be ignored for eval_proc itself.
+                # _load_initial_params re-applies the same overrides after
+                # ``prepare_global()`` clears gd, so the gd value stays in
+                # sync with the cached path.
+                apply_overrides(self.__defs, self.__gui_defaults)
+
                 # Creation of the python evaluation process for loading of the complete test
                 eval_proc = eval_process_init(api_request, 10, test_dir)
                 eval_proc.start()

src/testium/interpreter/test_items/dialog_choices_files/choices_dialog.py

@@ -221,6 +221,11 @@ def main(args, conn=None):
 
     if conn:
         settings.setValue(SettingsLastChoices, result)
+        # Flush before sending: the parent terminates this subprocess as soon
+        # as it reads the result, so the QSettings destructor never runs and
+        # the write would race the kill (lost under Flatpak — see the
+        # tested-references dialog for the full rationale).
+        settings.sync()
         conn.send([result, success])
         conn.close()
     else:

src/testium/interpreter/test_items/item_actions/__init__.py

@@ -5,6 +5,13 @@ from interpreter.test_items.item_actions.action import TestItemAction
 
 
 class TestItemActions(TestItem):
+    # Declarative action registry: subclasses set ``ACTIONS = {yaml_key: class}``
+    # as a class attribute (mirroring ``PARAMS``). It is read here to populate
+    # the runtime registry, and read identically by the schema export — no
+    # instantiation or source inspection required. ``register_actions()`` stays
+    # available as an imperative escape hatch for dynamic/conditional cases.
+    ACTIONS = {}
+
     def __init__(
         self, item_type, dict_actions, parent=None, status_queue=None, filename=""
     ):
@@ -12,7 +19,7 @@ class TestItemActions(TestItem):
         super().__init__(dict_actions, parent, status_queue, filename=filename)
         self._type = item_type
         self.is_container = False
-        self.action_classes = {}
+        self.action_classes = dict(type(self).ACTIONS)
         self.actions_token = None
         self.actions = []
         try:
@@ -24,6 +31,9 @@ class TestItemActions(TestItem):
             )
 
     def register_actions(self, **args: TestItemAction):
+        # Imperative escape hatch. The declarative ``ACTIONS`` class attribute
+        # covers every current subclass; use this only to add actions that
+        # can't be known at class-definition time (e.g. platform-conditional).
         for action_name, action_class in args.items():
             self.action_classes.update({action_name: action_class})
 

src/testium/interpreter/test_items/test_item.py

@@ -5,6 +5,9 @@ from copy import deepcopy
 from interpreter.test_items.test_result import TestResult, TestValue
 import api.testium as tm
 from interpreter.utils.params import TestItemParams
+from interpreter.utils.param_decl import (
+    Param, ParamSet, LIST, BLOCK, unknown_keys, missing_required,
+)
 from interpreter.utils.constants import TestItemType as cst_type
 from interpreter.utils.eval import eval_to_boolean, evaluate, post_evaluate
 from runtime.tum_except import ETUMSyntaxError, item_load_context
@@ -13,6 +16,32 @@ LOG_TEST_STOP = '<----- step "{}" finished'
 LOG_TEST_START = '-----> step "{}" started'
 
 
+# Parameters accepted by every test item, regardless of its type. Subclasses
+# concatenate their own ``PARAMS`` to this set; the merged result drives
+# unknown-param warnings and (later) the LSP schema export.
+COMMON_PARAMS = ParamSet(
+    Param("name",            doc="Display name shown in the GUI tree and reports."),
+    Param("doc",             doc="Free-form documentation; surfaced in tooltips."),
+    Param("skipped",         doc="If truthy, the step is skipped (static expression, "
+                                 "evaluated at load time)."),
+    Param("key",             doc="Report key used to classify the result "
+                                 "(typically <test>_PASS or <test>_FAIL)."),
+    Param("stop_on_failure", doc="If true, abort the surrounding container on failure."),
+    Param("execute_on_stop", doc="If true, run this step even when its container "
+                                 "is being stopped (cleanup)."),
+    Param("process_result",  doc="Post-evaluation expression applied to the test result."),
+    Param("store_result",    doc="Global-dict key in which to store the test result."),
+    Param("expected_result", doc="Expected outcome; the step is failed if it doesn't match."),
+    Param("no_fail",         doc="If truthy, never report a FAILURE for this step."),
+    Param("report",          doc="Per-step reporting override."),
+    Param("condition",       doc="Optional gating expression evaluated before each "
+                                 "run; false ⇒ the step is skipped."),
+    Param("steps",  kind=LIST, doc="Children (for container items)."),
+    Param("seq_filename",    doc="(internal) source .tum file of this step; injected "
+                                 "by the loader."),
+)
+
+
 class TestItem:
     pass
 
@@ -97,6 +126,11 @@ def test_data(item: TestItem, child: dict) -> dict:
 
 
 class TestItem:
+    # Subclasses override with their own ParamSet to opt into the declarative
+    # validation. While ``PARAMS`` is empty / unset, the base class skips the
+    # unknown-param check for this item type — keeps the migration incremental.
+    PARAMS = None
+
     def __init__(
         self, dict_item: dict = None, parent: TestItem = None, status_queue=None, filename = ""
     ):
@@ -134,6 +168,13 @@ class TestItem:
             # creation of the params object
             self._prms = TestItemParams(dict_item, parent)
 
+            # Declarative-params validation. Only kicks in when the concrete
+            # subclass declares ``PARAMS`` — items not yet migrated stay
+            # silent. Warnings (not errors) during the migration window so
+            # existing .tum files don't break suddenly; will be flipped to
+            # errors once every item has migrated.
+            self._validate_declared_params(dict_item)
+
             # getting parameters for the test item
             try:
                 self._name = self._prms.getParam("name", default="", processed=True)
@@ -190,6 +231,36 @@ class TestItem:
 
         self.result = TestResult(self, TestValue.FAILURE, "Failure by default")
 
+    def _validate_declared_params(self, dict_item):
+        """Warn on unknown / missing-required params, if PARAMS is declared.
+
+        The check is opt-in per subclass: it only runs when the concrete
+        class sets a non-empty ``PARAMS`` attribute. Items not yet migrated
+        produce no diagnostics — preserving the historical "silently accept
+        anything" behavior until they get their declaration.
+        """
+        if not self.PARAMS:
+            return
+        # ``self._type`` is the parent root type at this point (subclasses set
+        # it after super().__init__), so use the class name as a stable label
+        # in diagnostics. ``self._name`` was preset to the type name by every
+        # subclass before super() ran, which gives a useful prefix.
+        label = f"{type(self).__name__} '{self._name}'"
+        declared = COMMON_PARAMS + self.PARAMS
+        unknown = unknown_keys(declared, dict_item)
+        if unknown:
+            accepted = ", ".join(sorted(declared.names()))
+            for k in unknown:
+                tm.print_warn(
+                    f"{label}: unknown parameter '{k}'. Accepted: {accepted}."
+                )
+        missing = missing_required(declared, dict_item)
+        for k in missing:
+            raise ETUMSyntaxError(
+                f"{label}: required parameter '{k}' is missing.",
+                self._seq_filename,
+            )
+
     def _filter_dict_item(self, dict_item):
         # Stores the content of the step to be displayed
         # in the GUI

src/testium/interpreter/test_items/test_item_check.py

@@ -5,11 +5,22 @@ from runtime.tum_except import ETUMSyntaxError, item_load_context
 import api.testium as tm
 from interpreter.utils.constants import TestItemType as cst
 from interpreter.utils.eval import evaluate
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 
 class TestItemCheckValue(TestItem):
     """check item usage.
     check usage:{check: {name: check my func output, steps: ['$(pfn_echo) < 5']}}
     """
+
+    PARAMS = ParamSet(
+        Param("values", kind=LIST, required=True,
+              doc="List of expressions to evaluate. Each is expanded then "
+                  "evaluated; non-truthy results fail the check."),
+        # 'steps' is intentionally not redeclared here — it's the deprecated
+        # alias of 'values' and is already accepted by COMMON_PARAMS for
+        # container items. A runtime warning is emitted when 'steps' is used.
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_CHECK.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_choices_dialog.py

@@ -2,11 +2,25 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, BLOCK
 from runtime.tum_except import item_load_context
 import api.testium as tm
 
 
 class TestItemChoicesDialog(TestItemDialogBase):
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Prompt shown above the list of choices."),
+        Param("choices", kind=BLOCK, required=True,
+              doc="Tree of choices: either a list of strings, or a nested "
+                  "mapping {label: subchoices, ...} to build a multi-level menu."),
+        Param("icon", default=None,
+              doc="Default icon name shown next to each choice."),
+        Param("auto_result", default=None,
+              doc="Batch-mode selection (path or label). None ⇒ FAILURE."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_CHOICES_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_console.py

@@ -10,6 +10,7 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.item_actions import TestItemActions
 from interpreter.test_items.item_actions.action import TestItemAction
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from interpreter.test_items.test_result import TestResult, TestValue
 
 
@@ -21,6 +22,38 @@ class TestItemConsoleAction(TestItemAction):
 
 
 class TestItemConsoleOpen(TestItemConsoleAction):
+
+    PARAMS = ParamSet(
+        Param("protocol", required=True,
+              doc="Transport: 'telnet', 'ssh', 'rawtcp', 'serial' or 'terminal'."),
+        Param("write_delay", default=0,
+              doc="Inter-character write delay in ms (slow devices)."),
+        Param("log", doc="Path to a log file capturing the console traffic."),
+        Param("overwrite_log", default=True,
+              doc="If true, truncate the log file at open; else append."),
+        # telnet
+        Param("telnet_host", doc="Hostname/IP for the telnet target."),
+        Param("telnet_port", default=69, doc="TCP port for telnet."),
+        # ssh
+        Param("ssh_host", doc="Hostname/IP for the SSH target."),
+        Param("ssh_user", doc="SSH login user."),
+        Param("ssh_pwd",  doc="SSH password (if key-based auth is not used)."),
+        # rawtcp
+        Param("tcp_host", doc="Hostname/IP for a raw-TCP connection."),
+        Param("tcp_port", doc="TCP port for a raw-TCP connection."),
+        # serial
+        Param("serial_port",     doc="Serial device path (e.g. /dev/ttyUSB0 or COM3)."),
+        Param("serial_baudrate", doc="Serial baudrate."),
+        Param("buffered", default=True,
+              doc="If true, the serial console buffers received bytes between reads."),
+        # terminal
+        Param("terminal_path",
+              doc="Working directory for the local terminal protocol."),
+        Param("shell",
+              doc="Shell command used for the local terminal protocol "
+                  "(default: 'cmd.exe' on Windows, '/usr/bin/env bash' elsewhere)."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -283,6 +316,17 @@ class TestItemConsoleWriteLn(TestItemConsoleAction):
 
 
 class TestItemConsoleReadUntil(TestItemConsoleAction):
+
+    PARAMS = ParamSet(
+        Param("expected", required=True,
+              doc="Regex matched against incoming console output until found "
+                  "or until timeout."),
+        Param("timeout", default=-1,
+              doc="Seconds before giving up. Negative means infinite."),
+        Param("mute", default=False,
+              doc="If true, don't echo received bytes to testium's stdout/log."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -336,18 +380,27 @@ class TestItemConsoleReadUntil(TestItemConsoleAction):
 
 
 class TestItemConsole(TestItemActions):
+
+    PARAMS = ParamSet(
+        Param("console_name", required=True,
+              doc="Identifier of the console — used by every nested action to "
+                  "reach back the same transport. Multiple consoles can coexist "
+                  "as long as their names differ."),
+    )
+
+    ACTIONS = {
+        "open": TestItemConsoleOpen,
+        "close": TestItemConsoleClose,
+        "write": TestItemConsoleWrite,
+        "writeln": TestItemConsoleWriteLn,
+        "read_until": TestItemConsoleReadUntil,
+    }
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         super().__init__(
             cst.TYPE_CONSOLE, dict_item, parent, status_queue, filename=filename
         )
 
-        self.register_actions(
-            open=TestItemConsoleOpen,
-            close=TestItemConsoleClose,
-            write=TestItemConsoleWrite,
-            writeln=TestItemConsoleWriteLn,
-            read_until=TestItemConsoleReadUntil,
-        )
         self.actions_token = {}
 
         global console

src/testium/interpreter/test_items/test_item_cycle.py

@@ -8,9 +8,36 @@ from interpreter.test_items.test_result import TestResult, TestValue
 import api.testium as tm
 from interpreter.utils.params import TestItemParams
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, BLOCK
+
+
+# Sub-block validation: 'cycle' accepts an 'exit_condition:' mapping whose
+# own params are reported here so unknown keys inside it can be flagged
+# during a future Block-aware diagnostic pass. For now the parent only
+# declares that 'exit_condition' is an accepted top-level key.
+EXIT_CONDITION_PARAMS = ParamSet(
+    Param("time",      doc="HH:MM time of day after which the loop exits."),
+    Param("value",     doc="Expression; when truthy the loop exits."),
+    Param("file",      doc="Python file containing the exit-condition function."),
+    Param("func_name", doc="Function name in 'file' returning the exit value."),
+    Param("param",     doc="Arguments passed to the exit function."),
+    Param("eval", default="",
+          doc="Post-evaluation expression applied to the function's return."),
+)
 
 
 class TestItemCycle(TestItem):
+
+    PARAMS = ParamSet(
+        Param("iterator",
+              doc="Iterable (or string expanding to one) driving the loop. "
+                  "The current value is exposed as $(loop_param)."),
+        Param("exit_condition", kind=BLOCK,
+              doc="Optional block stopping the loop early: combine 'time', "
+                  "'value', or a 'file'+'func_name' pair (with optional "
+                  "'param' and 'eval')."),
+    )
+
     def __init__(self, dict_cycle, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_CYCLE.item_name
         super().__init__(dict_cycle, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_git.py

@@ -1,6 +1,7 @@
 from interpreter.test_items.test_item import (TestItem, test_run)
 from interpreter.test_items.test_result import (TestValue)
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 from runtime.tum_except import ETUMParamError, ETUMSyntaxError
 import interpreter.utils.version as git
 
@@ -8,6 +9,13 @@ class TestItemGit(TestItem):
     """
     This item expect only one parameter which is a string or list of string being the path to the git folder
     """
+
+    PARAMS = ParamSet(
+        Param("repo", kind=LIST, required=True,
+              doc="Path to a git checkout, or list of such paths. Each is "
+                  "reported with its current version (tag + dirty state)."),
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_GIT.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_group.py

@@ -1,10 +1,17 @@
 from interpreter.test_items.test_item import (TestItem, test_run)
 from interpreter.test_items.test_result import (TestResult, TestValue)
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import ParamSet
 from runtime.tum_except import ETUMSyntaxError
 import api.testium as tm
 
 class TestItemGroup(TestItem):
+
+    # 'group' has no item-specific parameters; 'steps' is handled by COMMON_PARAMS.
+    # Declaring an empty ParamSet still opts in to unknown-param validation
+    # (e.g. typo 'stop_on_failures').
+    PARAMS = ParamSet()
+
     def __init__(self, dict_cycle, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_GROUP.item_name
         super().__init__(dict_cycle, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_image_dialog.py

@@ -4,6 +4,7 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import item_load_context
 import api.testium as tm
 
@@ -12,6 +13,17 @@ class TestItemImageDialog(TestItemDialogBase):
     """dialog_image item usage.
     dialog_image name: Nice image, question: could you press the red button, filename: img.jpg
     """
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Prompt shown above the image."),
+        Param("filename", required=True,
+              doc="Path to the image file (relative to the test directory or absolute)."),
+        Param("auto_result", default=None,
+              doc="Outcome used in batch/non-interactive mode. Truthy ⇒ SUCCESS, "
+                  "None ⇒ FAILURE."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_IMAGE_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_json_rpc/__init__.py

@@ -11,6 +11,7 @@ from interpreter.test_items.item_actions.action import TestItemAction
 
 from interpreter.utils.constants import TestItemType as cst
 from interpreter.utils.eval import evaluate
+from interpreter.utils.param_decl import Param, ParamSet, BLOCK
 
 from interpreter.test_items.test_item_json_rpc.jsonrpc_adapters import (
     JrpcAdapter,
@@ -76,6 +77,20 @@ class TestItemJSRPCActionClose(TestItemAction):
 
 class TestItemJSRPCActionQuery(TestItemAction):
 
+    PARAMS = ParamSet(
+        Param("method", required=True,
+              doc="JSON-RPC method name to call."),
+        Param("params",
+              doc="Parameters payload (list, dict or scalar) sent to the method."),
+        Param("id", default="rand",
+              doc="JSON-RPC request id. 'rand' (default) ⇒ a random integer is used."),
+        Param("no_wait", default=False,
+              doc="If true, send the request without waiting for a response."),
+        Param("timeout", default=None,
+              doc="Seconds to wait for a response. None ⇒ inherits the transport "
+                  "default."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -129,6 +144,13 @@ class TestItemJSRPCActionQuery(TestItemAction):
 
 class TestItemJSRPCActionReceive(TestItemAction):
 
+    PARAMS = ParamSet(
+        Param("id", required=True,
+              doc="JSON-RPC request id whose response we expect."),
+        Param("timeout", default=None,
+              doc="Seconds to wait for the response. None ⇒ transport default."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -172,6 +194,29 @@ class TestItemJSON_RPC(TestItemActions):
     This item TBD
     """
 
+    PARAMS = ParamSet(
+        Param("console", kind=BLOCK,
+              doc="Console-transport block: {console_name, …}. Either 'console' "
+                  "or 'udp' must be set."),
+        Param("udp", kind=BLOCK,
+              doc="UDP-transport block: {host, port, …}. Either 'console' or "
+                  "'udp' must be set."),
+        Param("version", default="1.0",
+              doc="JSON-RPC protocol version ('1.0' or '2.0')."),
+        Param("timeout", required=True,
+              doc="Default seconds to wait for a JSON-RPC response across all "
+                  "child query/receive actions."),
+        Param("mute", default=False,
+              doc="If true, don't echo wire traffic to the log."),
+    )
+
+    ACTIONS = {
+        "open": TestItemJSRPCActionOpen,
+        "close": TestItemJSRPCActionClose,
+        "query": TestItemJSRPCActionQuery,
+        "receive": TestItemJSRPCActionReceive,
+    }
+
     def __init__(
         self, dict_item: dict, parent: TestItem = None, status_queue=None, filename=""
     ):
@@ -179,13 +224,6 @@ class TestItemJSON_RPC(TestItemActions):
             cst.TYPE_JSON_RPC, dict_item, parent, status_queue, filename=filename
         )
 
-        self.register_actions(
-            open=TestItemJSRPCActionOpen,
-            close=TestItemJSRPCActionClose,
-            query=TestItemJSRPCActionQuery,
-            receive=TestItemJSRPCActionReceive,
-        )
-
         # Console specific params
         self._console = self._prms.getParam("console", required=False)
         # UDP specific params

src/testium/interpreter/test_items/test_item_let.py

@@ -8,12 +8,20 @@ from interpreter.test_items.test_result import (TestResult, TestValue)
 from runtime.tum_except import ETUMSyntaxError, item_load_context
 import api.testium as tm
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 
 class TestItemLet(TestItem):
     """let item usage.
     let values: {variable1: a, variable2: /dev/ttyUSB0, variable3: 115200}
     let eval: {conditional_exec: "random.randint(1, 4)"}
     """
+
+    PARAMS = ParamSet(
+        Param("values", kind=LIST, required=True,
+              doc="Mapping (or list of single-pair mappings) of global-dict "
+                  "key → value to set. Values are expanded at execution time."),
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_LET.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_lua_func.py

@@ -11,6 +11,7 @@ import api.testium as tm
 from interpreter.utils.lua_func_exec import LuaFuncExecEngine
 from interpreter.utils.api_srv import api_request
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 
 _LUA_FUNC_CONTEXTS_KEY = "_lua_func_contexts"
 
@@ -21,6 +22,21 @@ class TestItemLuaFunc(TestItem):
     Optional: context_id: <id>  — share a persistent process with other lua_func items using the same id.
     """
 
+    PARAMS = ParamSet(
+        Param("file", required=True,
+              doc="Path to the .lua file containing the function."),
+        Param("func_name", required=True,
+              doc="Name of the function to call in the file."),
+        Param("param", kind=LIST,
+              doc="Arguments passed to the function. Each entry is expanded "
+                  "before the call. Special tokens $(loop_param) / $(loop_index) "
+                  "resolve from the surrounding cycle."),
+        Param("context_id", default=None,
+              doc="If set, the lua_func subprocess is kept alive and reused by "
+                  "every other lua_func item with the same context_id — enables "
+                  "shared in-memory state between successive calls."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_LUA_FUNCTION.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_msg_dialog.py

@@ -5,6 +5,7 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import item_load_context
 
 
@@ -12,6 +13,15 @@ class TestItemMsgDialog(TestItemDialogBase):
     """dialog_message item usage.
     dialog_message name: Nice message, question: Open the door and press OK
     """
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Message body shown to the user. Multi-line strings are supported."),
+        Param("auto_result", default=None,
+              doc="Outcome used in batch/non-interactive mode instead of waiting "
+                  "for the user. Truthy ⇒ SUCCESS, None ⇒ FAILURE."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_MESSAGE_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_note_dialog.py

@@ -2,11 +2,23 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import item_load_context
 import api.testium as tm
 
 
 class TestItemNoteDialog(TestItemDialogBase):
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Prompt shown above the note input field."),
+        Param("auto_result", default=None,
+              doc="Batch-mode outcome: None ⇒ FAILURE, 'cancel' ⇒ cancelled, "
+                  "any other truthy ⇒ SUCCESS with auto_value."),
+        Param("auto_value", default=None,
+              doc="Note text used in batch mode when auto_result is set."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_NOTE_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_parallel.py

@@ -6,6 +6,7 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestResult, TestValue
 from interpreter.utils.constants import TestItemType as cst
 from interpreter.utils.eval import eval_to_boolean
+from interpreter.utils.param_decl import Param, ParamSet, LIST, BLOCK, Enum
 from runtime.tum_except import ETUMSyntaxError
 from runtime.string_queue import StringQueue
 from runtime.stdout_redirect import stdio_redir
@@ -15,6 +16,12 @@ class TestItemParallelBranch(TestItemContainer):
     """One branch of a parallel item. Runs its children sequentially,
     optionally waiting for a condition before starting."""
 
+    PARAMS = ParamSet(
+        Param("wait_for", kind=BLOCK,
+              doc="Optional block {condition, timeout} that defers the branch "
+                  "start until the condition is truthy (or the timeout elapses)."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         super().__init__(cst.TYPE_PARALLEL_BRANCH, dict_item, parent, status_queue, filename=filename)
         self._wait_condition = None
@@ -87,6 +94,15 @@ class TestItemParallel(TestItemContainer):
                   - ...
     """
 
+    PARAMS = ParamSet(
+        Param("branches", kind=LIST, required=True,
+              doc="List of branch blocks (each branch holds its own 'steps' "
+                  "and optional 'wait_for')."),
+        Param("sync", kind=Enum("all", "any"), default="all",
+              doc="'all' (default) waits for every branch; 'any' returns as "
+                  "soon as the first branch completes."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         branches = dict_item.get("branches", [])
         if not branches:

src/testium/interpreter/test_items/test_item_py_func.py

@@ -11,6 +11,7 @@ import api.testium as tm
 from interpreter.utils.py_func_exec import PyFuncExecEngine
 from interpreter.utils.api_srv import api_request
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 
 _PY_FUNC_CONTEXTS_KEY = "_py_func_contexts"
 
@@ -21,6 +22,21 @@ class TestItemPyFunc(TestItem):
     Optional: context_id: <id>  — share a persistent process with other py_func items using the same id.
     """
 
+    PARAMS = ParamSet(
+        Param("file", required=True,
+              doc="Path to the .py file containing the function."),
+        Param("func_name", required=True,
+              doc="Name of the function to call in the file."),
+        Param("param", kind=LIST,
+              doc="Arguments passed to the function. Each entry is expanded "
+                  "before the call. Special tokens $(loop_param) / $(loop_index) "
+                  "resolve from the surrounding cycle."),
+        Param("context_id", default=None,
+              doc="If set, the py_func subprocess is kept alive and reused by "
+                  "every other py_func item with the same context_id — enables "
+                  "shared in-memory state between successive calls."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_PY_FUNCTION.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_question_dialog.py

@@ -2,6 +2,7 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import item_load_context
 
 
@@ -9,6 +10,14 @@ class TestItemQuestionDialog(TestItemDialogBase):
     """dialog_question item usage.
     dialog_question name: Nice question, question: "If OK, press OK, If not, press cancel"
     """
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Yes/No prompt presented to the user."),
+        Param("auto_result", default=None,
+              doc="Batch-mode answer ('yes'/'no' or truthy/falsy). None ⇒ FAILURE."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_QUESTION_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_report.py

@@ -3,9 +3,17 @@ from interpreter.test_items.test_item import (TestItem, test_run)
 from interpreter.test_items.test_result import (TestValue)
 from runtime.tum_except import ETUMSyntaxError
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 from interpreter.test_report.test_report import Export
 
 class TestItemReport(TestItem):
+
+    PARAMS = ParamSet(
+        Param("export", kind=LIST, required=True,
+              doc="List of exporters to run (junit, sqlite, …). Each entry is a "
+                  "mapping describing the exporter type and its parameters."),
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_REPORT.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_run.py

@@ -10,6 +10,7 @@ from interpreter.test_items.test_item import (TestItem, test_run)
 from interpreter.test_items.test_result import (TestValue)
 import api.testium as tm
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import ETUMSyntaxError, ETUMRuntimeError, item_load_context
 
 
@@ -24,9 +25,15 @@ def _testium_launch_cmd():
     appimage = os.environ.get("APPIMAGE")
     if appimage:
         return [appimage]
-    # Flatpak: re-launch via the Flatpak app id.
+    # Flatpak: re-launch via the Flatpak app id, but on the host side —
+    # the `flatpak` CLI cannot run inside our sandbox (no D-Bus access to the
+    # host Flatpak service, and the host binary would need host libs that are
+    # ABI-incompatible with the sandbox runtime). flatpak-spawn proxies the
+    # call to the host via org.freedesktop.Flatpak (allowed by --talk-name in
+    # the manifest).
     if os.path.isfile("/.flatpak-info"):
-        return ["flatpak", "run", "org.testium.Testium"]
+        return ["flatpak-spawn", "--host",
+                "flatpak", "run", "org.testium.Testium"]
     # PyInstaller frozen exe: sys.executable is the binary itself.
     if getattr(sys, "frozen", False):
         return [sys.executable]
@@ -51,6 +58,25 @@ def nowInBetween(start, end):
 
 
 class TestItemRun(TestItem):
+
+    PARAMS = ParamSet(
+        Param("tum", required=True,
+              doc="Path to the .tum file launched in a fresh testium instance."),
+        Param("param_file", default="",
+              doc="Optional path to a param.yaml passed to the sub-instance."),
+        Param("log_file", default="",
+              doc="Path where the sub-instance writes its log."),
+        Param("report_file", default="",
+              doc="Path where the sub-instance writes its report."),
+        Param("start_time",
+              doc="HH:MM time of day after which the sub-instance may run."),
+        Param("end_time",
+              doc="HH:MM time of day after which the sub-instance no longer runs."),
+        Param("wait_for_exec",
+              doc="If true, block until the time window opens. Requires both "
+                  "start_time and end_time."),
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_RUN.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_runtime_plot.py

@@ -11,6 +11,7 @@ from interpreter.test_items.item_actions import TestItemActions
 from interpreter.test_items.item_actions.action import TestItemAction
 from interpreter.utils.constants import TestItemType as cst
 from interpreter.utils.eval import evaluate
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 
 
 class TestItemPlotAction(TestItemAction):
@@ -21,6 +22,12 @@ class TestItemPlotAction(TestItemAction):
 
 
 class TestItemPlotActionOpen(TestItemPlotAction):
+
+    PARAMS = ParamSet(
+        Param("log_path", default=None,
+              doc="Optional file to which the plot data are appended."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -57,6 +64,15 @@ class TestItemPlotActionOpen(TestItemPlotAction):
 
 
 class TestItemPlotActionClose(TestItemPlotAction):
+
+    PARAMS = ParamSet(
+        Param("wait_dialog_exit", default=False,
+              doc="If true, the close action blocks until the user closes the "
+                  "plot window (or timeout)."),
+        Param("timeout", default=-1,
+              doc="Seconds to wait when wait_dialog_exit is true. Negative ⇒ infinite."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -96,6 +112,20 @@ class TestItemPlotActionClose(TestItemPlotAction):
 
 
 class TestItemPlotActionPeriodic(TestItemPlotAction):
+
+    PARAMS = ParamSet(
+        Param("period", required=True,
+              doc="Seconds between two calls of the periodic function."),
+        Param("file", required=True,
+              doc="Path to the .py file holding the periodic function."),
+        Param("func_name", required=True,
+              doc="Name of the periodic function."),
+        Param("param", kind=LIST,
+              doc="Arguments passed to the periodic function on each call."),
+        Param("eval", default="",
+              doc="Post-evaluation applied to the function's return value."),
+    )
+
     def __init__(
         self, action_name, dict_item, parent=None, status_queue=None, filename=""
     ):
@@ -169,6 +199,13 @@ class TestItemPlotActionAdd(TestItemPlotAction):
 
 
 class TestItemPlotActionLastValues(TestItemPlotAction):
+
+    PARAMS = ParamSet(
+        Param("name", kind=LIST,
+              doc="List of plot variable names whose last sample is returned. "
+                  "Result is stored in $(plv_<plot_name>) as a dict."),
+    )
+
     def __init__(self, action_name, dict_item, parent=None, status_queue=None, filename=""):
         super().__init__(
             action_name, cst.TYPE_GRAPH_ACTION, dict_item, parent, status_queue, filename=filename
@@ -219,18 +256,25 @@ class TestItemPlotActionExport(TestItemPlotAction):
 
 
 class TestItemPlot(TestItemActions):
+
+    PARAMS = ParamSet(
+        Param("plot_name", required=True,
+              doc="Identifier of the plot window — referenced by every nested "
+                  "action and by $(plv_<plot_name>) for last-values output."),
+    )
+
+    ACTIONS = {
+        "open": TestItemPlotActionOpen,
+        "close": TestItemPlotActionClose,
+        "periodic": TestItemPlotActionPeriodic,
+        "add": TestItemPlotActionAdd,
+        "last_value": TestItemPlotActionLastValues,
+        "export": TestItemPlotActionExport,
+    }
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         super().__init__(
             cst.TYPE_GRAPH, dict_item, parent, status_queue, filename=filename
         )
 
-        self.register_actions(
-            open=TestItemPlotActionOpen,
-            close=TestItemPlotActionClose,
-            periodic=TestItemPlotActionPeriodic,
-            add=TestItemPlotActionAdd,
-            last_value=TestItemPlotActionLastValues,
-            export=TestItemPlotActionExport,
-        )
-
         self.actions_token = self._prms.getParam("plot_name", required=True)

src/testium/interpreter/test_items/test_item_sleep.py

@@ -7,6 +7,7 @@ import api.testium as tm
 from interpreter.test_items.test_item import (TestItem, test_run)
 from interpreter.test_items.test_result import (TestValue)
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import ETUMSyntaxError, ETUMRuntimeError, item_load_context
 
 class TestItemSleep(TestItem):
@@ -14,6 +15,15 @@ class TestItemSleep(TestItem):
     sleep timeout: 10
     """
 
+    PARAMS = ParamSet(
+        Param("timeout", required=True,
+              doc="Duration to sleep. Number of seconds, or a string "
+                  "like '1d 2h 30m 15s'."),
+        Param("dialog", default=False,
+              doc="If true, show a cancel dialog (GUI mode) or an interactive "
+                  "Ctrl+C-able countdown (text mode)."),
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_SLEEP.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_tested_references.py

@@ -2,11 +2,23 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 from runtime.tum_except import item_load_context
 import api.testium as tm
 
 
 class TestItemTestedRefsDialog(TestItemDialogBase):
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Prompt asking the operator to enter the tested references."),
+        Param("reference", kind=LIST,
+              doc="Pre-filled list of references shown in the dialog."),
+        Param("auto_result", default=None,
+              doc="Batch-mode outcome: None ⇒ FAILURE, truthy ⇒ SUCCESS with "
+                  "the pre-filled references."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_REFERENCE_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_unittest.py

@@ -11,6 +11,7 @@ from interpreter.test_items.test_item import (TestItem, test_run, LOG_TEST_STOP,
 from interpreter.test_items.test_result import (TestResult, TestValue)
 from interpreter.test_items.test_item import test_data
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet, LIST
 from runtime.stdout_redirect import stdio_redir
 
 class UnittestResult(TextTestResult):
@@ -95,6 +96,15 @@ class TestItemUnittestElement(TestItem):
 
 
 class TestItemUnittestFile(TestItem):
+
+    PARAMS = ParamSet(
+        Param("test_file", required=True,
+              doc="Path to the Python unittest file (TestCase subclass)."),
+        Param("test_method", kind=LIST,
+              doc="Optional list of method names to restrict the run to. "
+                  "When empty, every test_* method in the file is run."),
+    )
+
     def __init__(self, dict_item, parent = None, status_queue=None, filename=""):
         self._name = cst.TYPE_UNITTEST.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/test_item_value_dialog.py

@@ -2,6 +2,7 @@ from interpreter.test_items.test_item import test_run
 from interpreter.test_items.test_result import TestValue
 from interpreter.test_items.test_item_dialog_base import TestItemDialogBase, _is_text_mode, _is_interactive
 from interpreter.utils.constants import TestItemType as cst
+from interpreter.utils.param_decl import Param, ParamSet
 from runtime.tum_except import item_load_context
 import api.testium as tm
 
@@ -10,6 +11,19 @@ class TestItemValueDialog(TestItemDialogBase):
     """dialog_value item usage.
     dialog_value name: Enter value, question: "Which value did you measure?"
     """
+
+    PARAMS = ParamSet(
+        Param("question", required=True,
+              doc="Prompt shown above the value input field."),
+        Param("default", default="",
+              doc="Pre-filled value of the input field."),
+        Param("auto_result", default=None,
+              doc="Batch-mode outcome: None ⇒ FAILURE, 'cancel' ⇒ cancelled, "
+                  "any other truthy ⇒ SUCCESS with auto_value."),
+        Param("auto_value", default=None,
+              doc="Value used in batch mode when auto_result is set."),
+    )
+
     def __init__(self, dict_item, parent=None, status_queue=None, filename=""):
         self._name = cst.TYPE_VALUE_DLG.item_name
         super().__init__(dict_item, parent, status_queue, filename=filename)

src/testium/interpreter/test_items/tested_references_files/tested_refs_dialog.py

@@ -76,6 +76,12 @@ def main(args, conn=None):
 
     if conn:
         settings.setValue(SettingsLastReference, result)
+        # Flush to disk *before* handing the result back: as soon as the parent
+        # receives it on the pipe it terminates this subprocess (SIGTERM, no
+        # handler), so the QSettings destructor never runs. Without sync() the
+        # write races the kill and is lost — reliably so under Flatpak, where
+        # the .conf is atomically renamed on the slower ~/.var/app overlay.
+        settings.sync()
         conn.send([result, success])
         conn.close()
     else:

src/testium/interpreter/test_set.py

@@ -29,6 +29,51 @@ def _build_item_path(item) -> str:
     return " > ".join(reversed(parts))
 
 
+def _flatten_actions(actions, out, parent_seq_name):
+    """Expand nested lists and included ``sequence`` entries into ``out`` as a
+    flat list of single test-item dicts, propagating each sequence's source
+    filename onto its items.
+
+    Replaces the previous approach, which spliced each entry into the step
+    list and rebuilt the whole list every time (O(n^2) over the step list, and
+    a rebuild that duplicated entries when a nested list held more than one
+    element). This single forward pass is linear.
+    """
+    for idx, action in enumerate(actions):
+        # a bare list raises its elements to the same level
+        if isinstance(action, (list, tuple)):
+            _flatten_actions(action, out, parent_seq_name)
+            continue
+        # a NoneType (e.g. pointing at an unused alias) contributes nothing
+        if action is None:
+            continue
+        # a 'sequence' (an included file) is spliced in, with its filename
+        # propagated onto each of its items
+        if isinstance(action, dict) and "sequence" in action:
+            sequence = action["sequence"]["data"]
+            f = action["sequence"]["filename"]
+            if isinstance(sequence, dict):
+                sequence = [{k: v} for k, v in sequence.items()]
+            # Case of an empty sequence
+            elif sequence is None:
+                tm.print_info(
+                    f"An empty sequence is loaded in '{parent_seq_name}'."
+                )
+                sequence = []
+            elif not isinstance(sequence, list):
+                raise ETUMSyntaxError(
+                    f"Syntax error in '{parent_seq_name}' step number {idx+1}. Sequence definition: '{str(action)}'",
+                    f
+                )
+            for s in sequence:
+                if isinstance(s, dict) and s:
+                    s[list(s.keys())[0]]["seq_filename"] = f
+            _flatten_actions(sequence, out, parent_seq_name)
+            continue
+
+        out.append(action)
+
+
 class TestSet:
     def __init__(
         self,
@@ -434,56 +479,16 @@ class TestSet:
                 f"No valid list of actions in sequence {parent_seq_name}",
                 file_name
             )
-        # first we merged to the same level 'sequence dict entries and list within the list
-        counter = 0
         test_dir = tm.gd("test_directory")
-        la = len(parent_seq_actions)
-        while counter < la:
-            action = parent_seq_actions[counter]
-            # if action is a list raise up to the the same level,
-            # ie insert action element into the parent_seq_actions
-            if isinstance(action, (list, tuple)):
-                parent_seq_actions[counter : counter + 1] = action
-                parent_seq_actions = (
-                    parent_seq_actions[:counter]
-                    + action
-                    + parent_seq_actions[counter + 1 :]
-                )
-                la = len(parent_seq_actions)
-                continue
-            # if action is a NoneType skip and continue
-            # (when pointing to an unused alias for instance)
-            if action is None:
-                counter += 1
-                continue
-            # if action is a sequence we insert its entry into the action list
-            if "sequence" in action:
-                sequence = action["sequence"]["data"]
-                f = action["sequence"]["filename"]
-                if isinstance(sequence, dict):
-                    sequence = [{k: v} for k, v in sequence.items()]
-                # Case of an empty sequence
-                elif sequence is None:
-                    tm.print_info(
-                        f"An empty sequence is loaded in '{parent_seq_name}'."
-                    )
-                    sequence = []
-                elif not isinstance(sequence, list):
-                    raise ETUMSyntaxError(
-                        f"Syntax error in '{parent_seq_name}' step number {counter+1}. Sequence definition: '{str(action)}'",
-                        f
-                    )
-                for s in sequence:
-                    s[list(s.keys())[0]]["seq_filename"] = f
-                parent_seq_actions = (
-                    parent_seq_actions[:counter]
-                    + sequence
-                    + parent_seq_actions[counter + 1 :]
-                )
-                la = len(parent_seq_actions)
-                continue
 
-            # Action is now for sure a list of dict of length 1
+        # Flatten nested lists and included 'sequence' entries to the same level
+        # in one linear pass (was an in-place splice + full list rebuild per
+        # entry: O(n^2) over the step list).
+        flat_actions = []
+        _flatten_actions(parent_seq_actions, flat_actions, parent_seq_name)
+
+        for action in flat_actions:
+            # Action is now for sure a dict of length 1
             k = list(action.keys())[0]
             if action[k].get("seq_filename", None) is None:
                 action[k]["seq_filename"] = file_name
@@ -546,8 +551,6 @@ class TestSet:
                     action[k]["seq_filename"]
                 )
 
-            counter += 1
-
         return ret
 
     def tree(self):

src/testium/interpreter/utils/bins.py

@@ -17,8 +17,12 @@ Public API
 ``reset()``             : clear the cache (mostly useful for tests)
 """
 
+import atexit
 import os
+import shlex
+import shutil
 import subprocess
+import tempfile
 
 import api.testium as tm
 from interpreter.utils.paths import sys_app_path_lin, sys_app_path_win
@@ -30,20 +34,6 @@ from runtime.tum_except import ETUMRuntimeError
 _PYTHON_CANDIDATES = ["python3", "python"]
 _LUA_CANDIDATES = ["lua", "lua5.5", "lua5.4", "lua5.3", "lua5.2", "lua5.1"]
 
-# When running inside a Flatpak, --filesystem=host-os mounts the host at
-# /run/host (read-only). Binaries and libraries from the host are not on the
-# sandbox PATH/LD_LIBRARY_PATH, so we probe and inject them explicitly.
-_FLATPAK_HOST_DIRS = [
-    "/run/host/usr/local/bin",
-    "/run/host/usr/bin",
-    "/run/host/bin",
-]
-_FLATPAK_HOST_LIB_DIRS = [
-    "/run/host/usr/lib",
-    "/run/host/usr/lib64",
-    "/run/host/usr/local/lib",
-]
-
 # Inside an AppImage, AppRun prepends $APPDIR/usr/bin to PATH and exports a
 # bundle-local PYTHONHOME / PYTHONPATH / LD_LIBRARY_PATH. We want py_func and
 # lua_func to run under the *host* interpreter (not the bundled one), so we
@@ -64,78 +54,183 @@ def _in_appimage():
     return "APPIMAGE" in os.environ
 
 
-def apply_host_lua_paths(env):
-    """Prepend host Lua module dirs to LUA_PATH / LUA_CPATH (Flatpak only).
-
-    Must be called after user-defined lua_env overrides are applied, so host
-    paths are always first regardless of user config. User-defined paths remain
-    in the variable but after the host ones.
-    """
-    if not _in_flatpak():
-        return
-    _LUA_VERSIONS = ["5.5", "5.4", "5.3", "5.2", "5.1"]
-    _HOST = "/run/host/usr"
-    cpath_dirs, lpath_dirs = [], []
-    for v in _LUA_VERSIONS:
-        for base in [f"{_HOST}/lib/lua/{v}",
-                     f"{_HOST}/lib64/lua/{v}",
-                     f"{_HOST}/lib/x86_64-linux-gnu/lua/{v}"]:
-            cpath_dirs.append(f"{base}/?.so")
-        lpath_dirs.append(f"{_HOST}/share/lua/{v}/?.lua")
-        lpath_dirs.append(f"{_HOST}/share/lua/{v}/?/init.lua")
-    sep = ";"
-    host_cpath = sep.join(cpath_dirs)
-    host_lpath = sep.join(lpath_dirs)
-    # ;; keeps Lua's compiled-in defaults at the end as last resort
-    env["LUA_CPATH"] = host_cpath + sep + env.get("LUA_CPATH", ";;")
-    env["LUA_PATH"]  = host_lpath + sep + env.get("LUA_PATH",  ";;")
-
-
 def apply_host_libs(env):
-    """Prepare *env* for launching a host binary from inside our bundle.
+    """Strip bundle-local entries from *env* so a host binary can run cleanly.
 
-    - Flatpak: prepend host library dirs to LD_LIBRARY_PATH so the dynamic
-      linker can find host .so files mounted under /run/host.
-    - AppImage: strip $APPDIR-prefixed entries from LD_LIBRARY_PATH and
-      PYTHONPATH and drop PYTHONHOME, so the host interpreter doesn't try
-      to load the bundled (incompatible) Python lib/site-packages.
-    - Otherwise: no-op.
+    Only meaningful for AppImage: removes $APPDIR-prefixed entries from
+    LD_LIBRARY_PATH / PYTHONPATH / PATH and drops PYTHONHOME, so the host
+    interpreter doesn't try to load the bundled (incompatible) Python
+    lib/site-packages. Flatpak is handled via flatpak-spawn --host instead
+    (see flatpak_host_spawn), so the sandbox env is irrelevant there.
     """
-    if _in_flatpak():
-        dirs = ":".join(d for d in _FLATPAK_HOST_LIB_DIRS if os.path.isdir(d))
-        if dirs:
-            existing = env.get("LD_LIBRARY_PATH", "")
-            env["LD_LIBRARY_PATH"] = dirs + (":" + existing if existing else "")
+    if not _in_appimage():
         return
-    if _in_appimage():
-        appdir = os.environ.get("APPDIR", "")
-        if appdir:
-            for var, sep in (("LD_LIBRARY_PATH", ":"),
-                             ("PYTHONPATH", os.pathsep),
-                             ("PATH", os.pathsep)):
-                cur = env.get(var, "")
-                if not cur:
-                    continue
-                cleaned = sep.join(
-                    p for p in cur.split(sep)
-                    if p and not p.startswith(appdir)
-                )
-                if cleaned:
-                    env[var] = cleaned
-                else:
-                    env.pop(var, None)
-        env.pop("PYTHONHOME", None)
+    appdir = os.environ.get("APPDIR", "")
+    if appdir:
+        for var, sep in (("LD_LIBRARY_PATH", ":"),
+                         ("PYTHONPATH", os.pathsep),
+                         ("PATH", os.pathsep)):
+            cur = env.get(var, "")
+            if not cur:
+                continue
+            cleaned = sep.join(
+                p for p in cur.split(sep)
+                if p and not p.startswith(appdir)
+            )
+            if cleaned:
+                env[var] = cleaned
+            else:
+                env.pop(var, None)
+    env.pop("PYTHONHOME", None)
+
+
+# ---------- Flatpak: spawn on host outside the sandbox -----------------------
+#
+# Inside a Flatpak the sandbox glibc is incompatible with host shared libraries,
+# so we can't run host Python/Lua under the sandbox runtime — `LD_LIBRARY_PATH`
+# tricks hit a `_dl_call_libc_early_init` assertion. The supported way out is
+# `flatpak-spawn --host`, which talks to the session-bus Flatpak D-Bus service
+# (org.freedesktop.Flatpak.Development) and asks it to spawn a process in the
+# host execution environment instead of inside our sandbox. The manifest must
+# grant `--talk-name=org.freedesktop.Flatpak` for the D-Bus call to be allowed.
+#
+# The host process can't see our /app/ contents (sandbox-only), so when we
+# spawn host Python/Lua to run `py_func` / `lua_func`, the cwd must be a
+# directory both sides can reach. /tmp is shared (--filesystem=/tmp), so we
+# stage the testium package there once per process and reuse it for every
+# spawn. In source mode (testium under $HOME) the host already sees the
+# original path, so we skip the copy.
+
+_staged_testium_path = None
+
+
+def _get_host_testium_path():
+    """Return a path to the testium package that the host can read.
+
+    - Source / wheel / PyInstaller install under $HOME → return testium_path()
+      as-is (host sees the same path via --filesystem=home).
+    - Flatpak bundle (testium under /app/) → stage a copy under /tmp on first
+      call and reuse it for the rest of the process.
+    """
+    global _staged_testium_path
+    if _staged_testium_path is not None:
+        return _staged_testium_path
+
+    # Imported lazily to avoid a circular import (paths.py -> api.testium).
+    from interpreter.utils.paths import testium_path
+    tp = testium_path()
+
+    if not tp.startswith("/app/"):
+        _staged_testium_path = tp
+        return tp
+
+    staged = tempfile.mkdtemp(prefix="testium_host_", dir="/tmp")
+    # copytree refuses to write into an existing dir unless dirs_exist_ok=True.
+    # mkdtemp creates the dir, so we copy *into* it.
+    for entry in os.listdir(tp):
+        src = os.path.join(tp, entry)
+        dst = os.path.join(staged, entry)
+        if os.path.isdir(src):
+            shutil.copytree(src, dst, symlinks=True)
+        else:
+            shutil.copy2(src, dst, follow_symlinks=False)
+    _staged_testium_path = staged
+    atexit.register(shutil.rmtree, staged, ignore_errors=True)
+    return staged
+
+
+_FORWARDED_ENV_KEYS = (
+    "HOME", "USER", "LOGNAME", "TMPDIR",
+    "XDG_RUNTIME_DIR", "XDG_DATA_HOME", "XDG_CONFIG_HOME", "XDG_CACHE_HOME",
+    "DBUS_SESSION_BUS_ADDRESS", "DISPLAY", "WAYLAND_DISPLAY",
+    "LANG", "LC_ALL",
+)
+
+
+def flatpak_host_spawn(interp_bin, cmd_args, host_cwd, extra_env=None):
+    """Build a flatpak-spawn --host command vector.
+
+    Args:
+        interp_bin: absolute path to the host interpreter (e.g. /usr/bin/python3).
+        cmd_args: list of arguments passed to the interpreter.
+        host_cwd: working directory on the host (must be reachable from host).
+        extra_env: optional {name: value} of env vars to set on the host side
+                   in addition to the default forwarded set. Values of ""
+                   unset the variable on the host.
+
+    Returns a list suitable for subprocess.Popen.
+    """
+    spawn = ["flatpak-spawn", "--host", f"--directory={host_cwd}"]
+    forwarded = {}
+    for key in _FORWARDED_ENV_KEYS:
+        val = os.environ.get(key)
+        if val:
+            forwarded[key] = val
+    if extra_env:
+        forwarded.update(extra_env)
+    for k, v in forwarded.items():
+        if v == "":
+            spawn.append(f"--unset-env={k}")
+        else:
+            spawn.append(f"--env={k}={v}")
+    spawn.append(interp_bin)
+    spawn.extend(cmd_args)
+    return spawn
+
+
+def host_console_command(shell_cmd, cwd):
+    """Build the argv to start *shell_cmd* as an ordinary interactive console.
+
+    *shell_cmd* is the command the caller chose (a string — shell-split — or
+    an argv list); the choice is preserved verbatim.
+
+    Outside Flatpak the command is returned unchanged. Inside Flatpak a bare
+    spawn would run in the sandbox under the runtime python3, so a host venv
+    (``/path/venv/bin/python3 -m mod``) can't see its pip deps. We simply run
+    it on the host with ``flatpak-spawn --host`` so it behaves like any other
+    terminal: flatpak-spawn passes the current environment through unchanged
+    and the shell (sourced venv, profile, …) sets things up as the user wants.
+    No env forwarding or scrubbing — the launcher's leaked PYTHONPATH points at
+    /app paths absent on the host, so it's inert there.
+    """
+    argv = shlex.split(shell_cmd) if isinstance(shell_cmd, str) else list(shell_cmd)
+    if not _in_flatpak():
+        return argv
+    return ["flatpak-spawn", "--host", f"--directory={cwd}", *argv]
+
+
+def _which_host_flatpak(name):
+    """Resolve a binary name (or absolute path) on the host via flatpak-spawn.
+
+    We can't probe /run/host/... because (a) only host-os is mounted there,
+    not arbitrary paths like /scratch, and (b) returning a /run/host path
+    would be useless — the host-side spawn sees a different filesystem and
+    needs the host-native path anyway.
+    """
+    if os.path.isabs(name):
+        cmd = flatpak_host_spawn("/bin/sh", ["-c", f'test -x "{name}"'],
+                                 host_cwd="/tmp")
+        try:
+            r = subprocess.run(cmd, capture_output=True, timeout=10)
+        except (FileNotFoundError, PermissionError, subprocess.TimeoutExpired):
+            return ""
+        return name if r.returncode == 0 else ""
+    cmd = flatpak_host_spawn("/bin/sh", ["-c", f'command -v "{name}"'],
+                             host_cwd="/tmp")
+    try:
+        r = subprocess.run(cmd, capture_output=True, text=True, timeout=10)
+    except (FileNotFoundError, PermissionError, subprocess.TimeoutExpired):
+        return ""
+    if r.returncode != 0:
+        return ""
+    return r.stdout.strip()
 
 
 def _which(name):
     if tm.OS() == "Windows":
         return sys_app_path_win(name)
     if _in_flatpak():
-        for d in _FLATPAK_HOST_DIRS:
-            p = os.path.join(d, name)
-            if os.path.isfile(p) and os.access(p, os.X_OK):
-                return p
-        return ""
+        return _which_host_flatpak(name)
     if _in_appimage():
         for d in _APPIMAGE_HOST_DIRS:
             p = os.path.join(d, name)
@@ -146,14 +241,33 @@ def _which(name):
 
 
 def _probe_env():
-    """Subprocess env for probing host binaries (adds host libs in Flatpak)."""
+    """Subprocess env for probing host binaries.
+
+    In AppImage we still need to scrub APPDIR-prefixed entries; in Flatpak we
+    delegate execution to the host via flatpak-spawn so the sandbox env doesn't
+    matter, but apply_host_libs is a no-op cost.
+    """
     env = os.environ.copy()
     apply_host_libs(env)
     return env
 
 
-def _python_version(path):
-    cmd = [path, "-c", "import sys; print(sys.version_info[:3])"]
+def _run_probe(cmd):
+    """Run a probe command, dispatching through flatpak-spawn --host in Flatpak.
+
+    Returns (stdout, stderr) as str, or None on failure.
+    """
+    if _in_flatpak():
+        spawn = flatpak_host_spawn(cmd[0], cmd[1:], host_cwd="/tmp")
+        try:
+            r = subprocess.run(
+                spawn, capture_output=True, text=True, timeout=10,
+            )
+        except (FileNotFoundError, PermissionError, subprocess.TimeoutExpired):
+            return None
+        if r.returncode != 0:
+            return None
+        return r.stdout, r.stderr
     try:
         r = subprocess.run(
             cmd, capture_output=True, text=True,
@@ -161,8 +275,15 @@ def _python_version(path):
         )
     except (FileNotFoundError, PermissionError, subprocess.TimeoutExpired):
         return None
+    return r.stdout, r.stderr
+
+
+def _python_version(path):
+    out = _run_probe([path, "-c", "import sys; print(sys.version_info[:3])"])
+    if out is None:
+        return None
     try:
-        return eval(r.stdout)
+        return eval(out[0])
     except Exception:
         return None
 
@@ -173,15 +294,11 @@ def _is_python3(path):
 
 
 def _lua_version(path):
-    try:
-        r = subprocess.run(
-            [path, "-v"], capture_output=True, text=True, timeout=10,
-            env=_probe_env(),
-        )
-    except (FileNotFoundError, PermissionError, subprocess.TimeoutExpired):
+    out = _run_probe([path, "-v"])
+    if out is None:
         return None
     # On Windows the version banner goes to stderr.
-    line = r.stdout or r.stderr
+    line = out[0] or out[1]
     try:
         major, minor, _patch = line.split(" ")[1].split(".")
         return (int(major), int(minor))
@@ -202,22 +319,33 @@ _SPECS = {
     "lua":    ("Lua 5.1+",  "lua_bin",    _LUA_CANDIDATES,    _is_lua51),
 }
 
+# Cached per (name, override) so that runtime changes to gd[gd_key] —
+# e.g. ``python_bin`` set from a YAML config file loaded *after*
+# eval_proc has already resolved its own interpreter — are picked up by
+# the next lookup instead of returning the stale, auto-discovered path.
+# Long-lived subprocesses (eval_proc) keep whatever they captured at
+# construction time, but every new PyProcessBase / FuncExecEngine spawned
+# afterwards sees the current override.
 _resolved = {}
 
 
 def _resolve(name):
-    if name in _resolved:
-        return _resolved[name]
-
     display, gd_key, candidates, validator = _SPECS[name]
     override = tm.gd(gd_key, "") or ""
 
+    cached = _resolved.get(name)
+    if cached is not None and cached[0] == override:
+        return cached[1]
+
     path = ""
     if override:
         # Absolute path: accept as-is (user knows exactly what they want).
         # Bare name: resolve via _which() so the override stays host-only in
         # Flatpak/AppImage instead of silently picking the bundled interpreter.
-        if os.path.isabs(override):
+        # In Flatpak we always defer to _which() so even absolute paths are
+        # checked from the host's perspective (the sandbox can't see e.g.
+        # /scratch/... paths that the user may have configured).
+        if os.path.isabs(override) and not _in_flatpak():
             resolved = override if (os.path.isfile(override)
                                     and os.access(override, os.X_OK)) else ""
         else:
@@ -239,7 +367,7 @@ def _resolve(name):
                 path = p
                 break
 
-    _resolved[name] = path
+    _resolved[name] = (override, path)
     return path
 
 

src/testium/interpreter/utils/include.py

@@ -6,10 +6,10 @@ from runtime.tum_except import ETUMFileError
 from interpreter.utils.template import template_to_test
 from copy import copy
 from interpreter.utils.globdict import global_dict
-from interpreter.utils.yaml_load import yaml_load
+from interpreter.utils.yaml_load import yaml_load, YAML_BASE_LOADER
 
 
-class TUMLoaderNoIncludes(yaml.Loader):
+class TUMLoaderNoIncludes(YAML_BASE_LOADER):
 
     def __init__(self, stream):
 

src/testium/interpreter/utils/lua_process.py

@@ -47,9 +47,16 @@ class LuaProcessBase:
         if self._process is not None:
             raise ETUMRuntimeError("The function subprocess has already been started.")
 
-        func_proc_path = os.path.realpath(
-            os.path.join(subproc_path(), "lua_func")
-        )
+        # In Flatpak the host can't see /app/lib/testium/lua_func, so use a
+        # staged copy under /tmp (shared between sandbox and host).
+        if bins._in_flatpak():
+            func_proc_path = os.path.join(
+                bins._get_host_testium_path(), "lua_func"
+            )
+        else:
+            func_proc_path = os.path.realpath(
+                os.path.join(subproc_path(), "lua_func")
+            )
 
         # POpen config
         CUST_ENV = {
@@ -71,7 +78,6 @@ class LuaProcessBase:
                     env[k] = e
                 else:
                     env[k] = e + ";" + env.get(k, "")
-        bins.apply_host_lua_paths(env)
 
         sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
         sock.bind(("localhost", 0))
@@ -79,8 +85,7 @@ class LuaProcessBase:
         sock.close()
 
         # POpen params
-        params = [
-            self._lbin,
+        cmd_args = [
             "main.lua",
             "--timeout",
             f"{self._timeout}",
@@ -91,14 +96,31 @@ class LuaProcessBase:
         ]
 
         if tm.debug_enabled() and tm.gd("debug_rpc", False):
-            params.append("--verbose")
+            cmd_args.append("--verbose")
+
+        if bins._in_flatpak():
+            # Run on the host outside the sandbox: avoids glibc ABI mismatches
+            # between the Flatpak runtime and host shared libraries.
+            host_env = {
+                k: env[k] for k in ("LUA_PATH", "LUA_CPATH", "PATH")
+                if k in env and env[k]
+            }
+            params = bins.flatpak_host_spawn(
+                self._lbin, cmd_args, host_cwd=func_proc_path,
+                extra_env=host_env,
+            )
+            popen_kwargs = {}
+        else:
+            params = [self._lbin, *cmd_args]
+            popen_kwargs = {"env": env, "cwd": func_proc_path}
 
         self._process = subprocess.Popen(
-            params, env=env, cwd=func_proc_path,
+            params,
             stdin=subprocess.DEVNULL,
             stdout=subprocess.PIPE,
             stderr=subprocess.PIPE,
             restore_signals=False,
+            **popen_kwargs,
         )
         # Route subprocess stdout/stderr (lua require failures, syntax
         # errors, anything written to fd 1/2 before the in-script

src/testium/interpreter/utils/param_decl.py

@@ -0,0 +1,175 @@
+"""Declarative description of a test item's accepted parameters.
+
+Each ``TestItem`` subclass declares its parameter surface as a class
+attribute::
+
+    class TestItemFoo(TestItem):
+        PARAMS = ParamSet(
+            Param("bar",      required=True,  doc="The bar value."),
+            Param("baz",      default=0,      doc="Optional baz."),
+            Param("modes",    kind=LIST,      doc="Iterable of modes."),
+            Param("strategy", kind=ENUM("a", "b"), doc="..."),
+            Param("opts",     kind=BLOCK,     doc="Sub-block."),
+        )
+
+The base ``TestItem.__init__`` consumes both ``COMMON_PARAMS`` (defined
+in ``test_item.py``) and the subclass ``PARAMS`` to:
+
+* warn on any key in the user's YAML that isn't declared anywhere
+  (catches typos like ``param_filee``);
+* expose a machine-readable schema for documentation generation and,
+  eventually, an LSP server.
+
+The descriptor is **purely about shape and naming**. Type coercion and
+runtime checking of expanded values remain the responsibility of each
+item's ``execute()`` method — most parameters are expressions
+(``$(...)`` / ``<| ... |>``) whose effective type is only known after
+expansion, so a static type would be misleading.
+
+Validation of *values* (e.g. ``start_time`` must match HH:MM) can be
+attached per-param via ``validate=lambda v: ...`` and is applied at
+execution time on the expanded value, not at load time.
+"""
+
+from dataclasses import dataclass, field
+from typing import Any, Callable, Optional, Union
+
+
+# ---------- Parameter "kinds" -------------------------------------------------
+#
+# These describe the YAML *shape* expected for a parameter, not its
+# semantic type. They drive the LSP completion (do we suggest a single
+# value, a list, a sub-block, an enum picker?) and the unknown-param
+# diagnostic; nothing more.
+
+SCALAR = "scalar"  # single value (string, number, bool, expression, ...)
+LIST = "list"      # YAML list — the historical ``getParamAll`` case
+BLOCK = "block"    # nested dict — e.g. ``cycle.exit:``
+
+
+@dataclass(frozen=True)
+class Enum:
+    """Closed enumeration of acceptable scalar values."""
+    values: tuple
+
+    def __init__(self, *values):
+        # frozen=True forbids assignment; bypass via object.__setattr__.
+        object.__setattr__(self, "values", tuple(values))
+
+    def __repr__(self):
+        return f"Enum({', '.join(repr(v) for v in self.values)})"
+
+
+Kind = Union[str, Enum]
+
+
+# ---------- The descriptor ----------------------------------------------------
+
+_MISSING = object()
+
+
+@dataclass(frozen=True)
+class Param:
+    """Declarative description of one accepted parameter.
+
+    Attributes
+    ----------
+    name : str
+        The YAML key.
+    kind : ``SCALAR`` (default) | ``LIST`` | ``BLOCK`` | ``Enum(...)``
+        The YAML shape expected.
+    required : bool
+        If True, missing the parameter is a load-time error.
+    default : Any
+        Default value when the parameter is absent. ``_MISSING`` when no
+        default was set (used to distinguish "absent" from "None").
+    doc : str
+        Free-form description used for hover / generated documentation.
+    validate : Optional[Callable[[Any], bool]]
+        Optional post-expansion validator, evaluated at ``execute()``
+        time on the effective (expanded) value. Returning ``False``
+        raises a clear error pointing at the param.
+    """
+    name: str
+    kind: Kind = SCALAR
+    required: bool = False
+    default: Any = _MISSING
+    doc: str = ""
+    validate: Optional[Callable[[Any], bool]] = None
+
+    def has_default(self):
+        return self.default is not _MISSING
+
+    def to_schema(self):
+        """Return a dict suitable for JSON Schema generation."""
+        s = {"name": self.name, "required": self.required, "doc": self.doc}
+        if isinstance(self.kind, Enum):
+            s["kind"] = "enum"
+            s["enum"] = list(self.kind.values)
+        else:
+            s["kind"] = self.kind
+        if self.has_default():
+            s["default"] = self.default
+        return s
+
+
+class ParamSet:
+    """Ordered, name-indexed collection of ``Param`` descriptors.
+
+    Supports concatenation (``COMMON_PARAMS + SUBCLASS_PARAMS``) to
+    merge the common surface with each item's own params. Later
+    declarations override earlier ones (so a subclass can tighten a
+    common param's docstring without redeclaring everything).
+    """
+
+    def __init__(self, *params):
+        self._params = {}
+        for p in params:
+            self.add(p)
+
+    def add(self, param):
+        if not isinstance(param, Param):
+            raise TypeError(f"ParamSet only accepts Param instances, got {type(param).__name__}")
+        self._params[param.name] = param
+
+    def __iter__(self):
+        return iter(self._params.values())
+
+    def __contains__(self, name):
+        return name in self._params
+
+    def __getitem__(self, name):
+        return self._params[name]
+
+    def names(self):
+        return tuple(self._params.keys())
+
+    def __add__(self, other):
+        if not isinstance(other, ParamSet):
+            return NotImplemented
+        merged = ParamSet()
+        merged._params = {**self._params, **other._params}
+        return merged
+
+    def to_schema(self):
+        return [p.to_schema() for p in self._params.values()]
+
+
+# ---------- Validation primitives --------------------------------------------
+
+def unknown_keys(declared, user_dict):
+    """Return the user-provided keys that are not declared in *declared*.
+
+    *declared* is a ``ParamSet``; *user_dict* is the raw YAML mapping
+    for the item. Unknown keys catch typos and obsolete parameters.
+    """
+    if not isinstance(user_dict, dict):
+        return ()
+    return tuple(k for k in user_dict.keys() if k not in declared)
+
+
+def missing_required(declared, user_dict):
+    """Return the names of declared required params absent from *user_dict*."""
+    if not isinstance(user_dict, dict):
+        return tuple(p.name for p in declared if p.required)
+    return tuple(p.name for p in declared if p.required and p.name not in user_dict)

src/testium/interpreter/utils/py_process.py

@@ -61,14 +61,18 @@ class PyProcessBase:
         if sock is not None:
             sock.close()
 
-        # Add the path of the subprocess (root sources of testium)
-        tstium_path = os.path.realpath(testium_path())
-        func_proc_path = os.path.realpath(subproc_path())
+        # In Flatpak the host can't see /app/lib/testium, so use a staged copy
+        # under /tmp (shared between sandbox and host) for both cwd and as the
+        # root in PYTHONPATH. Outside Flatpak the original paths are used.
+        if bins._in_flatpak():
+            tstium_path = bins._get_host_testium_path()
+            func_proc_path = tstium_path
+        else:
+            tstium_path = os.path.realpath(testium_path())
+            func_proc_path = os.path.realpath(subproc_path())
         env["PYTHONPATH"] = tstium_path + os.pathsep + self._ppath + os.pathsep + env.get("PYTHONPATH", "")
 
-        params = [
-            self._pbin,
-            # "-m",
+        cmd_args = [
             "py_func",
             "-p",
             f"{self._port}",
@@ -77,14 +81,31 @@ class PyProcessBase:
         ]
 
         if tm.debug_enabled() and tm.gd("debug_rpc", False):
-            params.append("-v")
+            cmd_args.append("-v")
+
+        if bins._in_flatpak():
+            # Run on the host outside the sandbox: avoids glibc ABI mismatches
+            # between the Flatpak runtime and host shared libraries.
+            host_env = {
+                k: env[k] for k in ("PYTHONPATH", "PATH")
+                if k in env and env[k]
+            }
+            params = bins.flatpak_host_spawn(
+                self._pbin, cmd_args, host_cwd=func_proc_path,
+                extra_env=host_env,
+            )
+            popen_kwargs = {}
+        else:
+            params = [self._pbin, *cmd_args]
+            popen_kwargs = {"env": env, "cwd": func_proc_path}
 
         self._process = subprocess.Popen(
-            params, env=env, cwd=func_proc_path,
+            params,
             stdin=subprocess.DEVNULL,
             stdout=subprocess.PIPE,
             stderr=subprocess.PIPE,
             restore_signals=False,
+            **popen_kwargs,
         )
         # Route subprocess stdout/stderr (early-startup errors,
         # unhandled exceptions, anything written to fd 1/2 before the

src/testium/interpreter/utils/template.py

@@ -1,33 +1,74 @@
+import io
 import os
 from sys import exc_info
-from jinja2 import Template
+from jinja2 import Environment
 from jinja2.exceptions import TemplateSyntaxError, TemplateError, UndefinedError
-from tempfile import TemporaryFile
 from interpreter.utils.yaml_load import print_yaml
 from runtime.tum_except import ETUMSyntaxError
 
 
+# One Environment reused for every render (default settings, i.e. identical
+# behaviour to jinja2.Template), plus a compiled-template cache so a file that
+# is included many times — or a test that is reloaded — is compiled only once.
+# Jinja compilation is the expensive step; render (variable substitution) stays
+# per-call. Cache is keyed on path + mtime + size so an edited file recompiles.
+_ENV = Environment()
+_template_cache = {}  # abspath -> (mtime_ns, size, compiled_template)
+
+
+class _RenderedStream(io.StringIO):
+    """A rendered template kept in memory.
+
+    Carries ``root`` (and ``name``) so the YAML loader resolves ``!include``
+    paths exactly as it did from the on-disk temp file this replaces — without
+    the write + seek + read round-trip (one temp file per included file). That
+    round-trip is pure overhead, and especially costly on slow storage.
+    """
+
+
+def _compiled_template(filename: str):
+    """Return the compiled jinja template for *filename*, reusing the cached
+    one when the file is unchanged (path + mtime + size)."""
+    key = os.path.abspath(filename)
+    try:
+        st = os.stat(filename)
+    except OSError:
+        st = None
+    if st is not None:
+        cached = _template_cache.get(key)
+        if (cached is not None
+                and cached[0] == st.st_mtime_ns
+                and cached[1] == st.st_size):
+            return cached[2]
+    with open(filename, "r") as f:
+        source = f.read()
+    template = _ENV.from_string(source)  # compile (may raise TemplateSyntaxError)
+    if st is not None:
+        _template_cache[key] = (st.st_mtime_ns, st.st_size, template)
+    return template
+
+
 def template_to_test(filename: str, params: list):
     """ Function which processes an eventual jinja2 template to a test file
     """
-    # Temporary file created to receive the processed include
-    # file
-    tmpf = TemporaryFile('w+t')
-    with open(filename, 'r') as f:
-        try:
-            j2_template = Template(f.read())
-        except TemplateError as e:
+    # Compile (cached) — a syntax error in the template surfaces here.
+    try:
+        j2_template = _compiled_template(filename)
+    except TemplateError as e:
+        with open(filename, "r") as f:
             print_yaml(f, filename)
-            type, value, tb = exc_info()
-            msg = "Template error"
-            if hasattr(value, 'lineno'):
-                msg = msg + f" on line {value.lineno}: "
-            else:
-                msg += ": "
-            raise ETUMSyntaxError(msg + str(e), filename)
+        type, value, tb = exc_info()
+        msg = "Template error"
+        if hasattr(value, 'lineno'):
+            msg = msg + f" on line {value.lineno}: "
+        else:
+            msg += ": "
+        raise ETUMSyntaxError(msg + str(e), filename)
+
+    # Render into memory (no temp file).
     try:
         params["include_directory"] = os.path.dirname(os.path.abspath(filename))
-        tmpf.write(j2_template.render(params))
+        rendered = j2_template.render(params)
     except TemplateSyntaxError as e:
         raise ETUMSyntaxError(f"""Template loading of file '{filename}' with following parameters '{str(params)}'
 Syntax error in template: {e.message}""")
@@ -42,8 +83,7 @@ Template rendering error: {e.message}""")
         raise ETUMSyntaxError(f"""Template loading of file '{filename}' with following parameters '{str(params)}'
 Unexpected error: {str(e)}""")
 
-    # return to begining of the temp file
-    tmpf.seek(0, os.SEEK_SET)
-    tmpf.root = os.path.dirname(filename)
-
-    return tmpf
+    stream = _RenderedStream(rendered)
+    stream.root = os.path.dirname(filename)
+    stream.name = filename
+    return stream

src/testium/interpreter/utils/test_init.py

@@ -11,7 +11,7 @@ import api.testium as tm
 import interpreter.utils.globdict as globdict
 import interpreter.utils.settings as prefs
 from interpreter.utils.paths import testium_path
-from interpreter.utils.yaml_load import yaml_load
+from interpreter.utils.yaml_load import yaml_load, YAML_BASE_LOADER
 from interpreter.utils import clear_recursively
 from runtime.tum_except import ETUMSyntaxError
 from interpreter.utils.params import expanse, eval_func_init
@@ -89,7 +89,7 @@ def locate_report_file(rep_file):
 def yamltodict(param_file, silent=True):
     # load of the file
     with open(param_file, "r") as fd:
-        dp = yaml_load(fd, param_file, yaml.Loader)
+        dp = yaml_load(fd, param_file, YAML_BASE_LOADER)
 
     if dp is None:
         tm.print_info(f"The YAML file '{param_file}' is empty.")
@@ -165,11 +165,14 @@ def env_init():
     _constants_init()
 
 
-def update_global(config_files, defines, gui_defaults, silent=False):
-    """Global dict updated with the content of the config file and a dict provided.
-    this function returns the resulting dict.
+def apply_overrides(defines, gui_defaults):
+    """Push GUI defaults then CLI defines into the global dict.
+
+    Extracted from update_global so it can be called *before* eval_proc
+    starts: interpreter overrides (python_bin, lua_bin) must be visible
+    to bins.python_bin() on its first lookup, which happens during
+    eval_process_init.
     """
-    # GUI preferences applied first
     for k, v in gui_defaults.items():
         try:
             val = ast.literal_eval(v)
@@ -177,7 +180,6 @@ def update_global(config_files, defines, gui_defaults, silent=False):
             val = v
         tm.setgd(k, val)
 
-    # Then command line defines
     for k, v in defines.items():
         try:
             val = ast.literal_eval(v)
@@ -185,6 +187,14 @@ def update_global(config_files, defines, gui_defaults, silent=False):
             val = v
         tm.setgd(k, val)
 
+
+def update_global(config_files, defines, gui_defaults, silent=False):
+    """Global dict updated with the content of the config file and a dict provided.
+    this function returns the resulting dict.
+    """
+    # GUI preferences applied first, then command line defines
+    apply_overrides(defines, gui_defaults)
+
     # Then the configuration files
     # load global dic before test item
     _feed_gd_with_params(config_files, silent)

src/testium/interpreter/utils/version.py

@@ -1,10 +1,54 @@
+import atexit
 import os
+import stat
 import sys
+import tempfile
 from importlib import import_module
 
 import interpreter.utils.settings as prefs
 import api.testium as tm
 
+
+# When running inside a Flatpak, the host /usr/bin/git is reachable at
+# /run/host/usr/bin/git but linked against host glibc/zlib, which the
+# sandbox can't load (``libz-ng.so.2`` not found). gitpython resolves git
+# eagerly on import and would crash the whole test run. We install a
+# tiny shell wrapper under /tmp that forwards to ``flatpak-spawn --host
+# git``, and point gitpython at it via ``GIT_PYTHON_GIT_EXECUTABLE``.
+_HOST_GIT_WRAPPER = None
+
+
+def _setup_flatpak_git():
+    global _HOST_GIT_WRAPPER
+    if not os.path.isfile("/.flatpak-info"):
+        return
+    if _HOST_GIT_WRAPPER is not None:
+        return
+    fd, path = tempfile.mkstemp(prefix="testium-git-host-", suffix=".sh", dir="/tmp")
+    with os.fdopen(fd, "w") as f:
+        f.write('#!/bin/sh\nexec flatpak-spawn --host git "$@"\n')
+    os.chmod(path, stat.S_IRWXU)
+    _HOST_GIT_WRAPPER = path
+    atexit.register(_cleanup_flatpak_git)
+    os.environ["GIT_PYTHON_GIT_EXECUTABLE"] = path
+    # Silence gitpython's warning if its refresh probe ever still fails;
+    # the wrapper itself should make the probe succeed.
+    os.environ.setdefault("GIT_PYTHON_REFRESH", "quiet")
+
+
+def _cleanup_flatpak_git():
+    global _HOST_GIT_WRAPPER
+    if _HOST_GIT_WRAPPER and os.path.isfile(_HOST_GIT_WRAPPER):
+        try:
+            os.unlink(_HOST_GIT_WRAPPER)
+        except OSError:
+            pass
+        _HOST_GIT_WRAPPER = None
+
+
+_setup_flatpak_git()
+
+
 _cached_versions = {}
 
 def repo_rev(path):

src/testium/interpreter/utils/yaml_load.py

@@ -1,3 +1,4 @@
+import yaml
 from yaml.parser import ParserError
 from yaml import load, Loader
 from yaml.scanner import ScannerError
@@ -5,6 +6,12 @@ from api.testium import print_debug
 from runtime.tum_except import ETUMSyntaxError
 import io
 
+# Use the libyaml-backed loader (much faster parsing) when PyYAML was built
+# with it, falling back to the pure-Python loader otherwise. The C loader
+# raises the same ParserError/ScannerError and supports the same custom
+# constructors (!include) and construct_* helpers the TUM loaders rely on.
+YAML_BASE_LOADER = yaml.CLoader if getattr(yaml, "__with_libyaml__", False) else yaml.Loader
+
 
 def print_yaml(file: io.TextIOWrapper, file_name):
     """ Prints YAML file if debug mode is activated.
@@ -21,10 +28,10 @@ def yaml_load(file, real_file_name: str, loader: Loader):
         return load(file, loader)
 
     except ParserError as e:
-        if isinstance(file, io.TextIOWrapper):
+        if isinstance(file, (io.TextIOWrapper, io.StringIO)):
             print_yaml(file, real_file_name)
         raise ETUMSyntaxError(f"yaml file parsing error: " + str(e), real_file_name)
     except ScannerError as e:
-        if isinstance(file, io.TextIOWrapper):
+        if isinstance(file, (io.TextIOWrapper, io.StringIO)):
             print_yaml(file, real_file_name)
         raise ETUMSyntaxError("yaml file scanning error: " + str(e), real_file_name)

src/testium/lsp/__init__.py

@@ -0,0 +1,16 @@
+"""testium language tooling.
+
+Hosts the JSON-Schema-style schema export of every test item type, and a
+``pygls`` language server that consumes the same schema to provide
+completion / hover / diagnostics for ``.tum`` files in any LSP-capable
+editor (VSCode, neovim, Helix, Emacs, …).
+
+Entry points (both surfaced through the ``testium`` CLI):
+
+- ``testium schema`` — dump the schema of every item type as JSON on stdout.
+  Zero runtime dependencies; can be used by editors that already speak the
+  YAML JSON Schema extension to get static completion immediately.
+
+- ``testium lsp`` — start the language server over stdio. Requires the
+  ``pygls`` optional dependency (``pip install testium[lsp]``).
+"""

src/testium/lsp/__main__.py

@@ -0,0 +1,6 @@
+"""Entry point for ``python -m testium.lsp`` (alternative to ``testium lsp``)."""
+
+from lsp.server import serve
+
+if __name__ == "__main__":
+    serve()

src/testium/lsp/schema.py

@@ -0,0 +1,122 @@
+"""Schema export of the test item registry.
+
+Walks every ``TestItemType`` entry (``interpreter/utils/constants.py``),
+combines its declared ``PARAMS`` with the common ones, and returns a
+serialisable structure keyed by ``item_cmd`` — the YAML key the user
+writes (e.g. ``sleep``, ``py_func``, ``dialog_message``).
+
+Items intentionally without ``PARAMS`` (the unstructured-body classes
+like console ``write``/``writeln`` or plot ``add``/``export``) are
+emitted as ``"params_declared": false`` so consumers know to suggest
+nothing for them rather than reporting a closed empty set.
+
+Action items (children of ``parallel``, ``console``, ``json_rpc``,
+``plot``) are registered separately under each parent's ``actions``
+entry — they're not top-level YAML keys, they live nested inside a
+parent's ``steps:``.
+"""
+
+import json
+
+from interpreter.utils.constants import TestItemType
+from interpreter.utils.test_init import _constants_init
+
+
+# Action class -> parent cmd (the action's parent in the YAML). Action classes
+# aren't first-class TestItemType entries (TYPE_*_ACTION is one generic bucket),
+# so we resolve their YAML key from the parent's declarative ``ACTIONS`` map.
+def _collect_action_classes(parent_class):
+    """Return {action_yaml_key: action_class} for a TestItemActions parent.
+
+    Each parent declares its actions as a class-level ``ACTIONS = {key: class}``
+    attribute (see ``item_actions/TestItemActions``). We read it directly — no
+    instantiation, no source inspection — so this works identically whether the
+    package runs from source, a wheel, or a frozen (PyInstaller) build where the
+    ``.py`` source isn't on disk.
+    """
+    return dict(getattr(parent_class, "ACTIONS", None) or {})
+
+
+def _params_to_schema(item_class, common_params):
+    """Return the params-portion of an item's schema entry.
+
+    Common params are flagged so consumers can render them differently
+    (an editor might show "common" parameters in a separate group).
+    """
+    own = getattr(item_class, "PARAMS", None)
+    if own is None:
+        return {"params_declared": False}
+    common_names = set(common_params.names())
+    params = []
+    for p in common_params:
+        d = p.to_schema()
+        d["common"] = True
+        params.append(d)
+    for p in own:
+        if p.name in common_names:
+            # Subclass overrode a common param (e.g. tightened doc).
+            for d in params:
+                if d["name"] == p.name:
+                    d.update(p.to_schema())
+                    break
+            continue
+        d = p.to_schema()
+        d["common"] = False
+        params.append(d)
+    return {"params_declared": True, "params": params}
+
+
+def dump_all_schemas():
+    """Return the full schema as a Python dict ready for json.dumps.
+
+    Shape:
+        {
+          "items": {
+            "sleep": {
+              "display_name": "Sleep",
+              "params_declared": true,
+              "params": [{name, kind, required, default?, doc, common}, ...],
+            },
+            "console": {
+              ...,
+              "actions": {"open": {...}, "close": {...}, ...},
+            },
+            ...
+          }
+        }
+    """
+    _constants_init()
+    # Imported lazily — pulls test_item.py which references constants.
+    from interpreter.test_items.test_item import COMMON_PARAMS
+
+    out = {"items": {}}
+    for tp in TestItemType:
+        cls = getattr(tp, "item_class", None)
+        if cls is None:
+            continue
+        # Action types (CONSOLE_ACTION, GRAPH_ACTION, JSON_RPC_ACTION) have no
+        # standalone YAML representation — skip them here, they show up under
+        # their parent's "actions" key.
+        cmd = tp.item_cmd
+        if cmd.endswith("_action"):
+            continue
+        entry = {"display_name": tp.item_name}
+        entry.update(_params_to_schema(cls, COMMON_PARAMS))
+
+        actions = _collect_action_classes(cls)
+        if actions:
+            entry["actions"] = {
+                name: _params_to_schema(acls, COMMON_PARAMS)
+                for name, acls in actions.items()
+            }
+            for name in entry["actions"]:
+                entry["actions"][name]["display_name"] = name
+
+        out["items"][cmd] = entry
+    return out
+
+
+def dump_all_schemas_json(indent=2):
+    """Same as ``dump_all_schemas`` but serialised to a JSON string."""
+    return json.dumps(dump_all_schemas(), indent=indent, sort_keys=False,
+                      default=str)

src/testium/lsp/server.py

@@ -0,0 +1,313 @@
+"""LSP server for ``.tum`` files.
+
+Features available so far:
+
+- **Completion** — when the user starts a new YAML step (``- <cursor>``),
+  the server proposes the full list of known item types. The completion
+  item carries a short hover-style description listing required and
+  optional parameters.
+- **Hover** — over a known item-type word (``sleep``, ``py_func``, …)
+  the server renders the same description in a popup.
+- **Document symbols (outline)** — every ``- <type>:`` line becomes an
+  entry in the editor's outline view. Nesting follows YAML indentation,
+  so containers (``group``, ``loop``, ``parallel``, ``console`` …)
+  display their children as a subtree.
+
+The server speaks LSP over stdio. Start it with::
+
+    testium lsp
+
+Editors invoke it through their LSP client; the connection layer
+(``vscode-languageclient``, ``nvim-lspconfig``, ``lsp-mode``, …) takes
+care of the JSON-RPC framing.
+
+Architecture notes
+------------------
+
+The schema is built once at server start (``dump_all_schemas()``) and
+kept in memory; an editor restart picks up upstream changes. The schema
+is the **only** source of truth — when testium adds a new item type or
+parameter, the LSP automatically exposes it without any change here.
+
+The current handlers stay deliberately heuristic on the parser side:
+completion uses a line-prefix regex, outline a per-line ``- <known>:``
+sweep with indentation tracking. A proper YAML+Jinja parsing pass is
+still pending and is the prerequisite for *parameter*-level completion
+and diagnostics.
+"""
+
+import re
+
+try:
+    # pygls 2.x moved LanguageServer under pygls.lsp.server. We pin >=1.3 in
+    # the optional dependency to stay open to either family, but the import
+    # path differs — try the new one first, then the legacy one.
+    try:
+        from pygls.lsp.server import LanguageServer
+    except ImportError:
+        from pygls.server import LanguageServer  # pygls < 2
+    from lsprotocol.types import (
+        TEXT_DOCUMENT_COMPLETION,
+        TEXT_DOCUMENT_DOCUMENT_SYMBOL,
+        TEXT_DOCUMENT_HOVER,
+        CompletionItem,
+        CompletionItemKind,
+        CompletionList,
+        CompletionOptions,
+        CompletionParams,
+        DocumentSymbol,
+        DocumentSymbolParams,
+        Hover,
+        HoverParams,
+        InsertTextFormat,
+        MarkupContent,
+        MarkupKind,
+        Position,
+        Range,
+        SymbolKind,
+    )
+except ImportError as exc:
+    # Surfaced by the CLI dispatcher with a friendly install hint.
+    raise
+
+
+from lsp.schema import dump_all_schemas
+
+
+_LINE_START_STEP = re.compile(r"^\s*-\s*([A-Za-z_][A-Za-z0-9_]*)?\s*:?\s*$")
+
+# Matches "- <identifier>:" for outline / hover purposes. Captures the start
+# column of the identifier and the identifier itself. Trailing tokens after
+# the colon (inline-form params, comments) are tolerated.
+_STEP_LINE = re.compile(r"^(?P<lead>\s*-\s*)(?P<ident>[A-Za-z_][A-Za-z0-9_]*)\s*:")
+
+# Matches a ``name: <value>`` line under an item — used by the outline pass
+# to surface the user's display name next to the item type.
+_NAME_FIELD = re.compile(r"^\s*name\s*:\s*(?P<value>.+?)\s*$")
+
+# Word boundary used by hover to extract the identifier under the cursor.
+_IDENT_AT = re.compile(r"[A-Za-z_][A-Za-z0-9_]*")
+
+
+def _render_item_markdown(cmd, entry):
+    """Render an item-type's schema entry as a Markdown hover string.
+
+    Reused by both the completion-item documentation and the hover
+    handler so the editor presents identical information regardless of
+    how the user reached it.
+    """
+    detail = entry.get("display_name", cmd)
+    lines = [f"**{cmd}** — {detail}", ""]
+    if entry.get("params_declared"):
+        non_common = [p for p in entry["params"] if not p["common"]]
+        required = [p for p in non_common if p["required"]]
+        optional = [p for p in non_common if not p["required"]]
+        if required:
+            lines.append("Required parameters:")
+            for p in required:
+                lines.append(f"- `{p['name']}` — {p['doc']}")
+            lines.append("")
+        if optional:
+            lines.append("Optional parameters:")
+            for p in optional:
+                lines.append(f"- `{p['name']}` — {p['doc']}")
+    else:
+        lines.append("(Parameter list is not described — this item's body is the "
+                     "raw user value.)")
+    return "\n".join(lines)
+
+
+def _build_item_completions(schema):
+    """Return a list of CompletionItem covering every top-level item type.
+
+    Each completion inserts ``<name>:`` with the cursor positioned after
+    the colon so the user can immediately start typing parameters.
+    """
+    items = []
+    for cmd, entry in schema["items"].items():
+        if cmd == "default":
+            # Root sentinel; never appears as a YAML key.
+            continue
+        items.append(
+            CompletionItem(
+                label=cmd,
+                kind=CompletionItemKind.Class,
+                detail=entry.get("display_name", cmd),
+                documentation=MarkupContent(
+                    kind=MarkupKind.Markdown,
+                    value=_render_item_markdown(cmd, entry),
+                ),
+                insert_text=f"{cmd}:",
+                insert_text_format=InsertTextFormat.PlainText,
+            )
+        )
+    items.sort(key=lambda it: it.label)
+    return items
+
+
+def _word_at(line, character):
+    """Return ``(start, end, text)`` of the identifier under ``character``.
+
+    Returns ``None`` when the cursor isn't on a word. Used by hover.
+    """
+    for m in _IDENT_AT.finditer(line):
+        if m.start() <= character <= m.end():
+            return m.start(), m.end(), m.group(0)
+    return None
+
+
+def _build_document_symbols(lines, item_cmds):
+    """Walk ``lines`` and produce a nested ``DocumentSymbol`` tree.
+
+    Heuristics (no YAML parsing yet):
+    - Each ``- <known_cmd>:`` line becomes a symbol.
+    - Nesting follows the indentation of the leading ``-``: a deeper-
+      indented step is treated as a child of the most recent shallower
+      step.
+    - The symbol's ``detail`` is the ``name: <value>`` field if found
+      within a small window after the step header (no YAML parsing —
+      we just look at indented lines that aren't another ``- …`` step).
+
+    The result is suitable for the LSP outline panel even when the
+    surrounding YAML is mid-edit and structurally invalid.
+    """
+    root_children = []
+    # Each stack entry: (indent_col, children_list_to_append_to,
+    #                    pending_parent_symbol or None).
+    stack = [(-1, root_children, None)]
+
+    def _attach_name(parent_symbol, start_line):
+        """Look for the nearest ``name:`` field in the children of ``parent``."""
+        if parent_symbol is None or start_line + 1 >= len(lines):
+            return
+        base_indent = len(lines[start_line]) - len(lines[start_line].lstrip(" "))
+        for j in range(start_line + 1, min(start_line + 10, len(lines))):
+            l = lines[j]
+            stripped = l.lstrip(" ")
+            indent = len(l) - len(stripped)
+            if indent <= base_indent and stripped.strip() != "":
+                break
+            m = _NAME_FIELD.match(l)
+            if m:
+                value = m.group("value").strip("\"' ")
+                parent_symbol.detail = value
+                return
+
+    for i, raw_line in enumerate(lines):
+        m = _STEP_LINE.match(raw_line)
+        if not m:
+            continue
+        cmd = m.group("ident")
+        if cmd not in item_cmds:
+            continue
+        indent = len(m.group("lead")) - len(m.group("lead").lstrip(" "))
+        # Pop the stack until we find a parent with strictly smaller indent.
+        while stack and stack[-1][0] >= indent:
+            stack.pop()
+        if not stack:
+            stack.append((-1, root_children, None))
+        parent_children = stack[-1][1]
+
+        ident_start = m.start("ident")
+        ident_end = m.end("ident")
+        symbol = DocumentSymbol(
+            name=cmd,
+            detail=None,
+            kind=SymbolKind.Function,
+            range=Range(
+                start=Position(line=i, character=0),
+                end=Position(line=i, character=len(raw_line.rstrip("\n"))),
+            ),
+            selection_range=Range(
+                start=Position(line=i, character=ident_start),
+                end=Position(line=i, character=ident_end),
+            ),
+            children=[],
+        )
+        parent_children.append(symbol)
+        stack.append((indent, symbol.children, symbol))
+        _attach_name(symbol, i)
+    return root_children
+
+
+def _make_server():
+    server = LanguageServer("testium-lsp", "0.1.0")
+    schema = dump_all_schemas()
+    item_completions = _build_item_completions(schema)
+    # Set of cmd names accepted by the outline / hover passes. We include
+    # action names (console open/close/…, plot open/close/…, …) too so they
+    # appear in the outline tree and respond to hover.
+    item_cmds = set()
+    for cmd, entry in schema["items"].items():
+        if cmd == "default":
+            continue
+        item_cmds.add(cmd)
+        item_cmds.update(entry.get("actions", {}).keys())
+
+    @server.feature(
+        TEXT_DOCUMENT_COMPLETION,
+        CompletionOptions(trigger_characters=["-", " "]),
+    )
+    def completion(params: CompletionParams):
+        doc = server.workspace.get_text_document(params.text_document.uri)
+        line_idx = params.position.line
+        if line_idx >= len(doc.lines):
+            return CompletionList(is_incomplete=False, items=[])
+        line = doc.lines[line_idx]
+        # Only look at what's left of the cursor.
+        prefix = line[: params.position.character]
+        if not _LINE_START_STEP.match(prefix):
+            return CompletionList(is_incomplete=False, items=[])
+        return CompletionList(is_incomplete=False, items=item_completions)
+
+    @server.feature(TEXT_DOCUMENT_HOVER)
+    def hover(params: HoverParams):
+        doc = server.workspace.get_text_document(params.text_document.uri)
+        line_idx = params.position.line
+        if line_idx >= len(doc.lines):
+            return None
+        line = doc.lines[line_idx]
+        # Only respond when the cursor is on the type part of a step line
+        # ("- sleep:") — never for arbitrary words in a string.
+        step_match = _STEP_LINE.match(line)
+        if not step_match:
+            return None
+        word = _word_at(line, params.position.character)
+        if word is None:
+            return None
+        start, end, text = word
+        if text != step_match.group("ident") or text not in item_cmds:
+            return None
+        # Resolve the entry: top-level item, or action of any parent.
+        entry = schema["items"].get(text)
+        if entry is None:
+            for parent_entry in schema["items"].values():
+                actions = parent_entry.get("actions") or {}
+                if text in actions:
+                    entry = actions[text]
+                    break
+        if entry is None:
+            return None
+        return Hover(
+            contents=MarkupContent(
+                kind=MarkupKind.Markdown,
+                value=_render_item_markdown(text, entry),
+            ),
+            range=Range(
+                start=Position(line=line_idx, character=start),
+                end=Position(line=line_idx, character=end),
+            ),
+        )
+
+    @server.feature(TEXT_DOCUMENT_DOCUMENT_SYMBOL)
+    def document_symbols(params: DocumentSymbolParams):
+        doc = server.workspace.get_text_document(params.text_document.uri)
+        return _build_document_symbols(doc.lines, item_cmds)
+
+    return server
+
+
+def serve():
+    """Start the LSP server on stdio. Blocks until the client disconnects."""
+    server = _make_server()
+    server.start_io()

src/testium/lua_func/json-rpc.lua

@@ -41,8 +41,7 @@ end
 --- INTERNAL: Handle requests from the client
 function JSONRPC:_handle_request(req)
     local method = self.methods[req.method]
-    local ok, ret
-    local res, err
+    local ok, ret, err
     if not method then
         if req.id then self:_send_error(req.id, string.format("Method '%s' not registered in lua server")) end
         return
@@ -52,15 +51,18 @@ function JSONRPC:_handle_request(req)
 
     -- Only send response if it's not a Notification (notifications have no ID)
     if req.id then
-        if ok then
-            res = ret
-            if res == nil then
-                self:_send_error(req.id, tostring(err))
-            else
-                self:_send({ jsonrpc = "2.0", result = { returned_value = res }, id = req.id })
-            end
-        else
+        if not ok then
+            -- pcall trapped a runtime error in the method itself.
+            self:_send_error(req.id, tostring(ret))
+        elseif err ~= nil then
+            -- Method ran but signaled a logical error via its 2nd return.
             self:_send_error(req.id, tostring(err))
+        else
+            -- Success. A user function returning nothing yields ret==nil;
+            -- encode it as JSON null so "returned_value" stays present.
+            local val = ret
+            if val == nil then val = json.null end
+            self:_send({ jsonrpc = "2.0", result = { returned_value = val }, id = req.id })
         end
     end
 end

src/testium/main_win/desktop_integration.py

@@ -0,0 +1,95 @@
+"""Install a desktop entry + icon under the user's data dir so desktop shells
+show the testium icon in the task bar / dock.
+
+On a native Wayland session GNOME takes a window's task-bar icon from the
+``.desktop`` file whose name (or ``StartupWMClass``) matches the window
+``app_id`` — ``QGuiApplication.setWindowIcon`` is ignored there. The portable
+channels (source checkout, PyInstaller binary, AppImage) install no system
+desktop file, so we drop an idempotent one in ``~/.local/share``. The window
+``app_id`` is set to ``testium`` (see ``QApplication.setDesktopFileName`` in
+``testium_win``), which is exactly this file's base name.
+
+Flatpak ships its own ``org.testium.Testium.desktop`` and keeps its own app id,
+so the caller skips this integration there.
+"""
+import os
+import sys
+
+from PySide6.QtCore import Qt
+from PySide6.QtGui import QPixmap
+
+# Must match QApplication.setDesktopFileName(...) for the GUI, and is used as
+# both the desktop-file base name and the StartupWMClass.
+APP_ID = "testium"
+
+
+def _launch_command():
+    """Best-effort Exec= for the menu entry. Not needed for icon matching, but
+    makes the entry actually launchable when possible."""
+    appimage = os.environ.get("APPIMAGE")
+    if appimage:
+        return f'"{appimage}"'
+    if getattr(sys, "frozen", False):
+        return f'"{os.path.abspath(sys.executable)}"'
+    argv0 = os.path.abspath(sys.argv[0]) if sys.argv and sys.argv[0] else ""
+    if argv0 and os.path.exists(argv0):
+        return f'"{os.path.abspath(sys.executable)}" "{argv0}"'
+    return f'"{os.path.abspath(sys.executable)}" -m testium'
+
+
+def ensure_desktop_entry():
+    """Create (or refresh) ~/.local/share icon + desktop entry. Best-effort:
+    any failure is swallowed so it can never take the GUI down.
+
+    Freedesktop-only: a no-op off Linux (Windows / macOS use the window icon)."""
+    if not sys.platform.startswith("linux"):
+        return
+    try:
+        data_home = os.environ.get("XDG_DATA_HOME") or os.path.join(
+            os.path.expanduser("~"), ".local", "share"
+        )
+        icon_dir = os.path.join(data_home, "icons", "hicolor", "256x256", "apps")
+        app_dir = os.path.join(data_home, "applications")
+        icon_path = os.path.join(icon_dir, f"{APP_ID}.png")
+        desktop_path = os.path.join(app_dir, f"{APP_ID}.desktop")
+
+        os.makedirs(icon_dir, exist_ok=True)
+        os.makedirs(app_dir, exist_ok=True)
+
+        # Icon: render the bundled Qt resource to a PNG once. Requires a live
+        # QGuiApplication (the caller creates it before calling us).
+        if not os.path.isfile(icon_path):
+            pixmap = QPixmap(u":/black/testium_logo.png")
+            if not pixmap.isNull():
+                pixmap = pixmap.scaled(
+                    256, 256,
+                    Qt.AspectRatioMode.KeepAspectRatio,
+                    Qt.TransformationMode.SmoothTransformation,
+                )
+                pixmap.save(icon_path, "PNG")
+
+        # Absolute Icon= path so the shell resolves it without an icon-cache
+        # refresh; StartupWMClass lets X11 / XWayland match too.
+        desktop = (
+            "[Desktop Entry]\n"
+            "Type=Application\n"
+            "Name=Testium\n"
+            "Comment=Test sequencer\n"
+            f"Icon={icon_path}\n"
+            f"Exec={_launch_command()} %f\n"
+            "Terminal=false\n"
+            f"StartupWMClass={APP_ID}\n"
+            "Categories=Utility;Development;\n"
+        )
+
+        # Write only when missing or changed, to avoid needless menu churn.
+        current = None
+        if os.path.isfile(desktop_path):
+            with open(desktop_path, "r") as fh:
+                current = fh.read()
+        if current != desktop:
+            with open(desktop_path, "w") as fh:
+                fh.write(desktop)
+    except Exception:
+        # Desktop integration is a nicety, never a hard requirement.
+        pass

src/testium/main_win/file_dialog.py

@@ -0,0 +1,21 @@
+"""Helpers for Qt file/directory dialogs.
+
+In Flatpak the native QFileDialog goes through the XDG document portal,
+which returns ``/run/user/UID/doc/.../<file>`` and only exposes the
+selected file — sibling files (param.yaml, scripts, recent paths in
+preferences, ...) are unreachable. Forcing Qt's own non-native dialog
+makes it walk the real filesystem mounted via ``--filesystem=home``
+and return a regular path.
+"""
+
+import os
+
+from PySide6.QtWidgets import QFileDialog
+
+
+def options():
+    """Default ``QFileDialog`` options for the current runtime."""
+    opts = QFileDialog.Options()
+    if os.path.isfile("/.flatpak-info"):
+        opts |= QFileDialog.Option.DontUseNativeDialog
+    return opts

src/testium/main_win/preference_win/preference_win.py

@@ -3,6 +3,7 @@ from PySide6.QtWidgets import QDialog, QFileDialog
 from PySide6.QtGui import QFont
 
 from main_win.preference_win.preference_core_win import Ui_preferenceWindow
+from main_win import file_dialog
 
 import interpreter.utils.settings as prefs
 
@@ -193,6 +194,7 @@ class PrefWindow(QDialog):
             self,
             caption="Select the default report directory",
             dir=self.ui.editDefaultReportPath.text(),
+            options=file_dialog.options(),
         )
         if path:
             self.ui.editDefaultReportPath.setText(path)
@@ -203,6 +205,7 @@ class PrefWindow(QDialog):
             self,
             caption="Select the default log directory",
             dir=self.ui.editDefaultLogPath.text(),
+            options=file_dialog.options(),
         )
         if path:
             self.ui.editDefaultLogPath.setText(path)
@@ -213,6 +216,7 @@ class PrefWindow(QDialog):
             self,
             caption="Select the python interpreter",
             dir=self.ui.editPythonPath.text(),
+            options=file_dialog.options(),
         )
         if path:
             self.ui.editPythonPath.setText(path)
@@ -220,7 +224,10 @@ class PrefWindow(QDialog):
     @Slot()
     def on_butLuaPath_pressed(self):
         path, _ = QFileDialog.getOpenFileName(
-            self, caption="Select the lua interpreter", dir=self.ui.editLuaPath.text()
+            self,
+            caption="Select the lua interpreter",
+            dir=self.ui.editLuaPath.text(),
+            options=file_dialog.options(),
         )
         if path:
             self.ui.editLuaPath.setText(path)

src/testium/main_win/test_file_manager.py

@@ -9,6 +9,7 @@ from PySide6.QtWidgets import QApplication, QFileDialog, QProgressDialog
 from interpreter.process import TestProcess
 from interpreter.utils.test_ctrl import TestSetController
 from main_win.test_controller_service import TestControllerService
+from main_win import file_dialog
 import interpreter.utils.settings as prefs
 from runtime.tum_except import ETUMFileError, ETUMRuntimeError
 
@@ -212,17 +213,9 @@ class TestFileManager:
         d = ""
         if w.testFile is not None:
             d = os.path.dirname(w.testFile)
-        # In Flatpak the native dialog goes through the XDG document portal,
-        # which returns /run/user/UID/doc/.../test.tum and only exposes the
-        # selected file — sibling files (param.yaml, .py, etc.) are unreachable.
-        # Force Qt's own dialog, which walks the real filesystem mounted via
-        # --filesystem=home and returns a regular path with sibling access.
-        options = QFileDialog.Options()
-        if os.path.isfile("/.flatpak-info"):
-            options |= QFileDialog.Option.DontUseNativeDialog
         file_name, _ = QFileDialog.getOpenFileName(
             w, "Open the test file", d,
-            "testium file (*.tum);;All Files (*)", options=options
+            "testium file (*.tum);;All Files (*)", options=file_dialog.options()
         )
         if file_name:
             self.reload(file_name)

src/testium/main_win/testium_win.py

@@ -37,6 +37,7 @@ from interpreter.utils.icons import icon_prefix
 
 from main_win.test_run.outlog import OutLog
 from main_win.test_run.test_run import ThreadTestStatus
+from main_win import file_dialog
 import interpreter.utils.settings as prefs
 from runtime.stdout_redirect import stdio_redir
 import api.testium as tm
@@ -484,7 +485,8 @@ class MainWindow(QMainWindow, Ui_MainWindow):
         else:
             initialPath = None
         fileName, _ = QFileDialog.getSaveFileName(
-            self, "Path to Log file", initialPath, "Log Files (*.log);;All Files (*)"
+            self, "Path to Log file", initialPath, "Log Files (*.log);;All Files (*)",
+            options=file_dialog.options(),
         )
         if fileName:
             shutil.copy(self.logFileName, fileName)
@@ -525,7 +527,8 @@ class MainWindow(QMainWindow, Ui_MainWindow):
         else:
             initialPath = None
         fileName, _ = QFileDialog.getSaveFileName(
-            self, "Path to log file", initialPath, "Log Files (*.log);;All Files (*)"
+            self, "Path to log file", initialPath, "Log Files (*.log);;All Files (*)",
+            options=file_dialog.options(),
         )
         if fileName:
             self.editLogFilePath.setText(fileName)
@@ -675,6 +678,24 @@ def MainWin(
     debug=False,
 ):
     app = QApplication(sys.argv)
+    # Application identity so desktop shells (GNOME, ...) show the testium
+    # icon in the task bar / dock instead of a generic one. On Wayland this
+    # sets the surface app_id; on X11/XWayland it sets WM_CLASS, so the window
+    # stops inheriting the launcher's class (e.g. "python3" under the AppImage,
+    # which is what GNOME was keying the wrong icon off) and the window icon
+    # below is used as the fallback. In Flatpak the id must be the Flatpak app
+    # id so it matches the installed desktop file.
+    app.setApplicationName("Testium")
+    app.setApplicationDisplayName("Testium")
+    app.setDesktopFileName(os.environ.get("FLATPAK_ID", "testium"))
+    app.setWindowIcon(QIcon(u":/black/testium_logo.png"))
+    # On native Wayland the task-bar icon comes from an installed desktop file
+    # matched to the app_id, not from setWindowIcon(). Flatpak ships its own;
+    # for the other Linux channels drop an idempotent one under ~/.local/share.
+    # Windows / macOS use the window icon set above, so this is Linux-only.
+    if sys.platform.startswith("linux") and not os.environ.get("FLATPAK_ID"):
+        from main_win.desktop_integration import ensure_desktop_entry
+        ensure_desktop_entry()
     ui = MainWindow(
         test_file,
         config_files,

test/benchmark/.gitignore

@@ -0,0 +1 @@
+cases/

test/benchmark/README.md

@@ -0,0 +1,116 @@
+# Load-time benchmark
+
+Measures how long *testium* takes to **load** a `.tum` test tree — template
+rendering (jinja) + YAML parsing + test-tree construction — *without* executing
+it. Purpose: get reproducible numbers before/after load-path optimisations, and
+attribute any gain to a specific part of the pipeline.
+
+It is meant for *very long* tests, the kind you can build with `jinja` loops and
+`!include`, where load time becomes noticeable.
+
+## Files
+
+| File | Role |
+|------|------|
+| `gen_bench_test.py` | Generates a synthetic `.tum` tree (the test input). |
+| `load_bench.py` | Drives the **real** loader in-process and times it. |
+| `run.sh` | Convenience: generate + time across profiles, using the project venv. |
+| `cases/` | Generated trees (git-ignored, recreated on demand). |
+
+The benchmark `.tum` files are **generated**, not committed — the generator is
+the artifact. They use only `let` leaves and `group` containers, so loading has
+no runtime side effect (no subprocess, no `<| |>` eval) and the timing reflects
+the parse/build pipeline alone.
+
+## Quick start
+
+```bash
+# default matrix (all profiles), 5 repeats each
+./test/benchmark/run.sh
+
+# one profile at one size
+./test/benchmark/run.sh repeat 2000
+
+# more repeats for a tighter min
+REPEAT=10 ./test/benchmark/run.sh includes 1000
+```
+
+`run.sh` uses the project venv at `test/tmp/.venv` (created by `./run.sh`). If it
+is missing, run `./run.sh` once first.
+
+To drive the harness directly on any `.tum` (not just generated ones):
+
+```bash
+test/tmp/.venv/bin/python3 test/benchmark/load_bench.py --repeat 5 --quiet path/to/main.tum
+```
+
+## Profiles
+
+Each profile isolates one cost. `--size` is the profile-specific count.
+
+| Profile | What it builds | Stresses |
+|---------|----------------|----------|
+| `flat` | one main file, N inline `let` steps | big YAML parse + linear object build |
+| `includes` | main `!include`s N **distinct** sub-files | per-include template+YAML+tempfile, `sequence` splice |
+| `repeat` | main `!include`s the **same** parametrised leaf N times | jinja **recompilation** of an identical template |
+| `jinja` | one main file, `{% for %}` emitting N steps | single large render + single large parse |
+| `deep` | nested includes, depth N | include recursion (see caveat) |
+| `mix` | groups + jinja loop + distinct + repeated includes | realistic blend |
+
+## Reading the output
+
+```
+phase              min      median
+initial         0.1131      0.1285   <- pass 1: discover config files (no includes)
+loadtest        1.0724      1.0900   <- config fixpoint loop + full recursive include load
+build           0.1850      0.1976   <- TestSet: load_test_recursively tree build
+total           1.3886      1.4227
+counters  (last run):
+  templates :    1003 calls   0.5247s  (exclusive: jinja compile+render+tempfile)
+  yaml      :    1004 parses   1.4696s  (inclusive of nested includes)
+```
+
+- **min** is the headline (least noisy); median is a sanity check.
+- **initial / loadtest / build** map to the three pipeline stages in
+  `interpreter/process.py` and `interpreter/test_set.py`. The main file is
+  rendered+parsed across `initial` *and* `loadtest` (the loader does ~3 passes).
+- **templates** = number of `template_to_test()` calls and their *exclusive*
+  wall time (one file render each — pure jinja compile+render+tempfile I/O).
+  A high count with the same source file = recompilation, the `repeat` case.
+- **yaml** = number of `yaml_load()` parses. Its time is *inclusive* of nested
+  includes, so use the **count** for attribution, not the seconds.
+
+## Mapping to the optimisation axes
+
+| Axis (see DESIGN / discussion) | Watch | Best profile to prove it |
+|--------------------------------|-------|--------------------------|
+| 1 — cache compiled jinja templates | `templates` time drops, count unchanged | `repeat` |
+| 2 — drop the tempfile round-trip | `templates` time drops | `includes`, `repeat`, `mix` |
+| 3 — C YAML loader (libyaml) | `yaml` time / `loadtest` drops | `flat`, `jinja` |
+| 6 — O(n²) sequence splice | `build` drops | `includes`, `mix` |
+
+## How to compare before/after a change
+
+1. Run the matrix on the current code, keep the output.
+2. Apply one axis.
+3. Re-run the **same** profiles/sizes; compare `min` per phase and the counters.
+
+Change one axis at a time so the attribution is clean. Run on an idle machine
+(and note the disk: on a USB stick the tempfile round-trip of axis 2 weighs
+more).
+
+## Caveat: deep includes
+
+The loader is recursive and spends ~10 stack frames per include level, so
+`deep` hits Python's `RecursionError` around ~90 nested levels. The harness
+reports this cleanly instead of crashing. Real tests are *wide* (many steps /
+many includes), not deep, so `includes`/`repeat`/`jinja`/`mix` are the
+representative "very long" cases.
+
+## Notes
+
+- No execution is triggered — timing stops where `Batch` would mark the test
+  *loaded*.
+- The profiles contain no `<| |>`, so the external eval process is not started.
+  Pass `--with-eval` to `load_bench.py` for trees that evaluate at load time.
+- Numbers are machine- and disk-specific; only compare runs from the same host.

test/benchmark/gen_bench_test.py

@@ -0,0 +1,179 @@
+#!/usr/bin/env python3
+"""Generate synthetic ``.tum`` test trees to benchmark *load* time.
+
+The generated trees are deliberately cheap to *build* (only ``let`` leaves and
+``group`` containers — no subprocess, no runtime side effect) so the load
+benchmark measures the parse / template / tree-build pipeline and nothing else.
+
+Profiles, each targeting a specific cost in the loader:
+
+  flat      one main file, N inline ``let`` steps, no include, no jinja.
+            Baseline: YAML parse of a big document + linear object build.
+
+  includes  main ``!include``s N *distinct* sub-files (a few steps each).
+            Stresses the per-include template+YAML+tempfile round-trip and the
+            ``sequence`` splice in test_set.load_test_recursively.
+
+  repeat    main ``!include``s the *same* parametrised leaf file N times.
+            Stresses jinja *recompilation*: the compiled template is identical
+            every time, only the render params (idx) differ -> the case a
+            template cache collapses.
+
+  jinja     one main file whose ``{% for %}`` loop emits N steps.
+            Stresses a single large jinja render + a single large YAML parse.
+
+  deep      nested includes, depth N (main -> d0 -> d1 -> ...).
+            Stresses include recursion and per-level template+YAML.
+
+  mix       a realistic blend: groups, a jinja loop, distinct includes and a
+            repeated parametrised include.
+
+Usage:
+    gen_bench_test.py --profile repeat --size 1000 --out cases/repeat_1000
+    -> writes <out>/main.tum (+ includes, + param.yaml) and prints the path.
+"""
+import argparse
+import os
+import shutil
+
+
+def _let(indent, i, name=None):
+    name = name if name is not None else f"s{i}"
+    pad = " " * indent
+    return (
+        f"{pad}- let:\n"
+        f"{pad}    name: {name}\n"
+        f"{pad}    values:\n"
+        f"{pad}        - k{i}: {i}\n"
+    )
+
+
+def gen_flat(out, n):
+    body = "".join(_let(8, i) for i in range(n))
+    main = f"main:\n    name: bench flat {n}\n    steps:\n{body}"
+    _write(out, "main.tum", main)
+
+
+def gen_includes(out, n):
+    steps = "".join(f"        - !include inc_{i}.tum\n" for i in range(n))
+    main = f"main:\n    name: bench includes {n}\n    steps:\n{steps}"
+    _write(out, "main.tum", main)
+    for i in range(n):
+        # each include is a YAML *sequence* (list of steps)
+        seq = "".join(_let(0, i * 3 + j, name=f"inc{i}_{j}") for j in range(3))
+        _write(out, f"inc_{i}.tum", seq)
+
+
+def gen_repeat(out, n):
+    steps = "".join(
+        f"        - !include {{file: leaf.tum, idx: {i}}}\n" for i in range(n)
+    )
+    main = f"main:\n    name: bench repeat {n}\n    steps:\n{steps}"
+    _write(out, "main.tum", main)
+    leaf = (
+        "- let:\n"
+        "    name: leaf_{{ idx }}\n"
+        "    values:\n"
+        "        - leaf_{{ idx }}: {{ idx }}\n"
+    )
+    _write(out, "leaf.tum", leaf)
+
+
+def gen_jinja(out, n):
+    main = (
+        f"main:\n    name: bench jinja {n}\n    steps:\n"
+        "{% for i in range(" + str(n) + ") %}\n"
+        "        - let:\n"
+        "            name: j{{ i }}\n"
+        "            values:\n"
+        "                - k{{ i }}: {{ i }}\n"
+        "{% endfor %}\n"
+    )
+    _write(out, "main.tum", main)
+
+
+def gen_deep(out, n):
+    main = (
+        f"main:\n    name: bench deep {n}\n    steps:\n"
+        "        - let:\n            name: top\n            values:\n                - a: 0\n"
+        "        - !include d_0.tum\n"
+    )
+    _write(out, "main.tum", main)
+    for i in range(n):
+        seq = _let(0, i, name=f"d{i}")
+        if i < n - 1:
+            seq += f"- !include d_{i + 1}.tum\n"
+        _write(out, f"d_{i}.tum", seq)
+
+
+def gen_mix(out, n):
+    # n groups, each: 2 inline lets, one distinct include, one repeated include,
+    # plus a small jinja loop. Roughly ~6*n steps.
+    per = max(1, n)
+    parts = [f"main:\n    name: bench mix {n}\n    steps:\n"]
+    for g in range(per):
+        parts.append(
+            f"        - group:\n"
+            f"            name: grp{g}\n"
+            f"            steps:\n"
+        )
+        parts.append(_let(16, g * 2, name=f"g{g}_a"))
+        parts.append(_let(16, g * 2 + 1, name=f"g{g}_b"))
+        parts.append(f"                - !include inc_{g}.tum\n")
+        parts.append(f"                - !include {{file: leaf.tum, idx: {g}}}\n")
+        parts.append(
+            "{% for i in range(3) %}\n"
+            f"                - let:\n"
+            f"                    name: g{g}_j{{{{ i }}}}\n"
+            f"                    values:\n"
+            f"                        - g{g}_k{{{{ i }}}}: {{{{ i }}}}\n"
+            "{% endfor %}\n"
+        )
+    _write(out, "main.tum", "".join(parts))
+    for g in range(per):
+        _write(out, f"inc_{g}.tum", _let(0, g, name=f"mixinc{g}"))
+    _write(
+        out,
+        "leaf.tum",
+        "- let:\n    name: mixleaf_{{ idx }}\n    values:\n        - mixleaf_{{ idx }}: {{ idx }}\n",
+    )
+
+
+PROFILES = {
+    "flat": gen_flat,
+    "includes": gen_includes,
+    "repeat": gen_repeat,
+    "jinja": gen_jinja,
+    "deep": gen_deep,
+    "mix": gen_mix,
+}
+
+
+def _write(out, name, content):
+    with open(os.path.join(out, name), "w") as f:
+        f.write(content)
+
+
+def main():
+    ap = argparse.ArgumentParser(description=__doc__,
+                                 formatter_class=argparse.RawDescriptionHelpFormatter)
+    ap.add_argument("--profile", required=True, choices=sorted(PROFILES))
+    ap.add_argument("--size", type=int, default=1000,
+                    help="profile-specific count (steps / includes / depth)")
+    ap.add_argument("--out", required=True, help="output directory (recreated)")
+    args = ap.parse_args()
+
+    out = os.path.abspath(args.out)
+    if os.path.isdir(out):
+        shutil.rmtree(out)
+    os.makedirs(out)
+
+    # minimal config file so the loader does not emit "no param file" noise
+    _write(out, "param.yaml", "bench_dummy: 1\n")
+
+    PROFILES[args.profile](out, args.size)
+    print(os.path.join(out, "main.tum"))
+
+
+if __name__ == "__main__":
+    main()

test/benchmark/load_bench.py

@@ -0,0 +1,200 @@
+#!/usr/bin/env python3
+"""Time the testium *load* pipeline on a given ``.tum`` tree.
+
+It drives the real loader code (``TestProcess._load_initial_params`` /
+``_load_test`` then ``TestSet(...)``) in-process, so the numbers track the
+production path and stay honest as the code evolves. Execution is never
+triggered — we stop exactly where ``Batch`` would report the test as *loaded*.
+
+Reported per run, over ``--repeat`` iterations (min is the headline, least
+noisy):
+
+    initial   first pass: discover config files (template+YAML, no includes)
+    loadtest  config-file fixpoint loop + full recursive include/template/YAML
+    build     TestSet construction: the load_test_recursively tree build
+    total     sum of the three
+
+Plus instrumentation counters (exact call counts, wall time) for the two
+hot leaves the optimisation axes target:
+
+    templates  jinja template_to_test() calls   (axis 1 compile cache, axis 2 tempfile)
+    yaml       yaml_load() parses               (axis 3 C loader)
+
+template time is exclusive (one file render); yaml time is wall-inclusive of
+nested includes, so lean on the *counts* for attribution.
+
+Must run inside the project venv (jinja2, pyyaml, telnetlib3, ...). The
+benchmark profiles contain no ``<| |>`` so the external eval process is not
+needed; pass --with-eval to start it for faithfulness on eval-heavy trees.
+
+Usage (see run.sh for the convenience wrapper):
+    test/tmp/.venv/bin/python3 test/benchmark/load_bench.py [--repeat 5] <main.tum>
+"""
+import argparse
+import os
+import statistics
+import sys
+from queue import Queue
+from time import perf_counter
+
+# --- bootstrap: src/testium for flat imports, src for `import testium` --------
+HERE = os.path.dirname(os.path.abspath(__file__))
+ROOT = os.path.abspath(os.path.join(HERE, "..", ".."))
+sys.path.insert(0, os.path.join(ROOT, "src"))
+sys.path.insert(0, os.path.join(ROOT, "src", "testium"))
+
+import api.testium as tm  # noqa: E402
+from interpreter.utils.test_init import env_init, apply_overrides  # noqa: E402
+from interpreter.utils.test_ctrl import TestSetController  # noqa: E402
+from interpreter.process import TestProcess  # noqa: E402
+from interpreter.test_set import TestSet  # noqa: E402
+from interpreter.utils.py_eval import eval_process_init  # noqa: E402
+from interpreter.utils.api_srv import api_request  # noqa: E402
+
+# --- instrumentation: count + time the two hot leaves -------------------------
+import interpreter.process as _proc  # noqa: E402
+import interpreter.utils.include as _inc  # noqa: E402
+import interpreter.utils.test_init as _ti  # noqa: E402
+import interpreter.utils.template as _tpl  # noqa: E402
+import interpreter.utils.yaml_load as _yl  # noqa: E402
+
+_C = {"tpl_n": 0, "tpl_t": 0.0, "yaml_n": 0, "yaml_t": 0.0}
+_orig_tpl = _tpl.template_to_test
+_orig_yaml = _yl.yaml_load
+
+
+def _wrap_tpl(*a, **k):
+    t = perf_counter()
+    try:
+        return _orig_tpl(*a, **k)
+    finally:
+        _C["tpl_t"] += perf_counter() - t
+        _C["tpl_n"] += 1
+
+
+def _wrap_yaml(*a, **k):
+    t = perf_counter()
+    try:
+        return _orig_yaml(*a, **k)
+    finally:
+        _C["yaml_t"] += perf_counter() - t
+        _C["yaml_n"] += 1
+
+
+# rebind in every module that did `from ... import template_to_test / yaml_load`
+for _m in (_proc, _inc):
+    _m.template_to_test = _wrap_tpl
+for _m in (_proc, _inc, _ti):
+    _m.yaml_load = _wrap_yaml
+
+
+def _reset_counters():
+    _C.update(tpl_n=0, tpl_t=0.0, yaml_n=0, yaml_t=0.0)
+
+
+def load_once(tp, fname, test_dir):
+    """One full load (no execution). Returns (initial, loadtest, build) seconds."""
+    t0 = perf_counter()
+    init_pf, gv = tp._load_initial_params(test_dir)
+    t1 = perf_counter()
+    test_dict, _pf = tp._load_test(init_pf, gv)
+    t2 = perf_counter()
+    TestSet(fname, test_dict, Queue())
+    t3 = perf_counter()
+    return (t1 - t0, t2 - t1, t3 - t2)
+
+
+def main():
+    ap = argparse.ArgumentParser(description=__doc__,
+                                 formatter_class=argparse.RawDescriptionHelpFormatter)
+    ap.add_argument("main_tum", help="path to the generated main.tum")
+    ap.add_argument("--repeat", type=int, default=5)
+    ap.add_argument("--with-eval", action="store_true",
+                    help="start the external eval process (needed only for <| |> at load)")
+    ap.add_argument("--quiet", action="store_true",
+                    help="silence the loader's INFO output during runs")
+    args = ap.parse_args()
+
+    fname = os.path.abspath(args.main_tum)
+    if not os.path.isfile(fname):
+        ap.error(f"not found: {fname}")
+    test_dir = os.path.dirname(fname)
+
+    env_init()
+    apply_overrides({}, {})
+
+    eval_proc = None
+    if args.with_eval:
+        eval_proc = eval_process_init(api_request, 10, test_dir)
+        eval_proc.start()
+        eval_proc.wait_ready(10)
+
+    if args.quiet:
+        # the loader prints a couple of INFO lines per config file; mute stdout
+        # around the measured section to avoid I/O skew.
+        devnull = open(os.devnull, "w")
+        real_stdout = sys.stdout
+
+    tp = TestProcess(fname, Queue(), TestSetController(),
+                     config_files=[], defines={}, gui_defaults={}, text_mode=True)
+
+    samples = []  # list of (initial, loadtest, build)
+    last_counters = None
+    try:
+        for r in range(args.repeat):
+            _reset_counters()
+            if args.quiet:
+                sys.stdout = devnull
+            try:
+                samples.append(load_once(tp, fname, test_dir))
+            except RecursionError:
+                if args.quiet:
+                    sys.stdout = real_stdout
+                print(f"file      : {fname}")
+                print("ERROR     : RecursionError during load — the include "
+                      "nesting is too deep for the recursive loader.\n"
+                      "            (each include level costs ~10 stack frames; "
+                      "raise sys.setrecursionlimit to probe further.)")
+                return 2
+            except Exception as e:  # noqa: BLE001 - report, don't crash the bench
+                if args.quiet:
+                    sys.stdout = real_stdout
+                print(f"file      : {fname}")
+                print(f"ERROR     : load failed: {type(e).__name__}: {e}")
+                return 2
+            finally:
+                if args.quiet:
+                    sys.stdout = real_stdout
+            last_counters = dict(_C)
+    finally:
+        if eval_proc is not None:
+            eval_proc.stop()
+            eval_proc.join()
+        if args.quiet:
+            devnull.close()
+
+    initial = [s[0] for s in samples]
+    loadtest = [s[1] for s in samples]
+    build = [s[2] for s in samples]
+    total = [sum(s) for s in samples]
+
+    def stat(xs):
+        return min(xs), statistics.median(xs)
+
+    print(f"file      : {fname}")
+    print(f"repeats   : {args.repeat}   (showing  min | median, seconds)")
+    print(f"{'phase':<10}{'min':>12}{'median':>12}")
+    for name, xs in (("initial", initial), ("loadtest", loadtest),
+                     ("build", build), ("total", total)):
+        mn, md = stat(xs)
+        print(f"{name:<10}{mn:>12.4f}{md:>12.4f}")
+    if last_counters:
+        print("counters  (last run):")
+        print(f"  templates : {last_counters['tpl_n']:>7d} calls   "
+              f"{last_counters['tpl_t']:>8.4f}s  (exclusive: jinja compile+render+tempfile)")
+        print(f"  yaml      : {last_counters['yaml_n']:>7d} parses  "
+              f"{last_counters['yaml_t']:>8.4f}s  (inclusive of nested includes)")
+
+
+if __name__ == "__main__":
+    sys.exit(main() or 0)

test/benchmark/run.sh

@@ -0,0 +1,49 @@
+#!/bin/bash
+# Load-time benchmark driver: generate synthetic .tum trees and time the
+# testium load pipeline on them, using the project venv.
+#
+# Usage:
+#   ./test/benchmark/run.sh                     # default matrix (all profiles)
+#   ./test/benchmark/run.sh <profile> <size>    # one profile at one size
+#   REPEAT=10 ./test/benchmark/run.sh repeat 2000
+#
+# Profiles: flat includes repeat jinja deep mix   (see gen_bench_test.py)
+#
+# Generated trees go under test/benchmark/cases/ (git-ignored). The numbers
+# are wall-clock; run on an otherwise idle machine and compare min values.
+set -e
+
+SCRIPT_DIR="$(realpath "$(dirname "$(readlink -f "$0")")")"
+PROJECT_DIR="$(realpath "$SCRIPT_DIR/../..")"
+VPY="$PROJECT_DIR/test/tmp/.venv/bin/python3"
+CASES="$SCRIPT_DIR/cases"
+REPEAT="${REPEAT:-5}"
+
+if [ ! -x "$VPY" ]; then
+    echo "ERROR: project venv not found at $VPY — run ./run.sh once to create it." >&2
+    exit 1
+fi
+
+bench() {
+    local profile="$1" size="$2"
+    local out="$CASES/${profile}_${size}"
+    local main
+    main="$("$VPY" "$SCRIPT_DIR/gen_bench_test.py" --profile "$profile" --size "$size" --out "$out")"
+    echo "===== profile=$profile size=$size ====="
+    "$VPY" "$SCRIPT_DIR/load_bench.py" --repeat "$REPEAT" --quiet "$main"
+    echo
+}
+
+if [ $# -eq 2 ]; then
+    bench "$1" "$2"
+    exit 0
+fi
+
+# Default matrix. 'deep' is kept small: the recursive loader hits Python's
+# recursion limit around ~90 nested include levels.
+bench flat     2000
+bench includes 1000
+bench repeat   1000
+bench jinja    2000
+bench deep     40
+bench mix      300

test/robustness/console serial/rsl_terminal_robustness.tum

@@ -1,69 +0,0 @@
-# Main
-################################################################################
-main:
-  name:                               Serial Terminal bug reproducer
-  version:                            0.1
-  steps:
-    - group:
-        name:                         Test preparation
-        steps:
-          - console:
-              name:             Open RSL Simulator Terminal
-              console_name:     RSL_simulator
-              steps:
-                  - open:
-                      protocol:   terminal
-                      terminal_path: $(rslsimulatorpath)
-                  - writeln: "pwd"
-                  - read_until: {expected: "$", timeout: 5}
-                  - writeln: "./RSverify $(rsTx)" # /dev/ttyMUE1
-                  - read_until: {expected: "RSL controller>", timeout: 5}
-                  - writeln: "setportconf 0 115200 none 8 1 1 255"
-                  - read_until: {expected: "RSL controller>", timeout: 5}
-                  - writeln: "send4ever 0 0"
-                  - read_until: {expected: "RSL controller>", timeout: 5}
-
-          - console:
-              name:                   Open the EUT console
-              console_name:           cons_target
-              doc:                    Initiates the console of the target in order
-                                      to be ready to capture its traces.
-              stop_on_failure:        True
-              steps:
-                - open:
-                    protocol:         serial
-                    serial_port:      $(rsRx) # /dev/ttyMUE2
-                    serial_baudrate:  115200
-
-    - loop:
-        name:                         Qualification loop
-        stop_on_failure:              False
-        steps:
-          - py_func:
-              name:                   Capture the RS serial output
-              file:                   $(test_directory)/terminal_bug_reproducer.py
-              func_name:              RetreiveData
-              param:
-                - cons_target
-
-          - sleep: {timeout: 1}
-
-# Cleanup sequence
-#-------------------------------------------------------------------------------
-    - group:
-        name:                         Cleanup
-        execute_on_stop:              True
-        steps:
-          - console:
-              name:                   Close the target console
-              console_name:           cons_target
-              execute_on_stop:        True
-              steps:
-                - close:
-
-          - console:
-              name:                   Close the RSL_simulator
-              console_name:           RSL_simulator
-              execute_on_stop:        True
-              steps:
-                - close:

test/robustness/console serial/terminal_bug_reproducer.py

@@ -1,26 +0,0 @@
-import api.testium as tm
-
-def RetreiveData(console_name):
-    print("--------------- retrieving data ---------------")
-    result = 0
-    cons   = tm.console(console_name)
-
-    if cons is None:
-        print("--------------- The console does not exist ---------------")
-    else:
-        try:
-            is_finished = False
-            while not is_finished:
-                status, d = cons.read_until('\n', timeout=0, return_data=True, mute=True)
-                if 0 == status:
-                    print("--------------- Data ---------------")
-                    print(d)
-                else:
-                    print("--------------- No data ---------------")
-                    print("Status: ", status)
-                    is_finished = True
-        except:
-            print("--------------- Error retrieving data ---------------")
-            result = -1
-
-    return result

test/robustness/console terminal/generate_char.sh

@@ -1,9 +0,0 @@
-chars='<=>| -,;:!/."()[]{}*\&#%+012345689abcdefghiklmnopqrstuvwxyzABCD'
-for j in {1..256} ;
-do
-    for i in {1..256} ; do
-        echo -n "${chars:RANDOM%${#chars}:1}"
-    done
-    echo
-    sleep 0.01
-done

test/robustness/console terminal/terminal_bug_reproducer.py

@@ -1,26 +0,0 @@
-import api.testium as tm
-
-def RetreiveData(console_name):
-    print("--------------- retrieving data ---------------")
-    result = 0
-    cons   = tm.console(console_name)
-
-    if cons is None:
-        print("--------------- The console does not exist ---------------")
-    else:
-        try:
-            is_finished = False
-            while not is_finished:
-                status, d = cons.read_until('\n', timeout=0, return_data=True, mute=True)
-                if 0 == status:
-                    print("--------------- Data ---------------")
-                    print(d)
-                else:
-                    print("--------------- No data ---------------")
-                    print("Status: ", status)
-                    is_finished = True
-        except:
-            print("--------------- Error retrieving data ---------------")
-            result = -1
-
-    return result

test/robustness/console terminal/terminal_robustness.tum

@@ -1,50 +0,0 @@
-# Main
-################################################################################
-main:
-  name:                               Terminal bug reproducer
-  version:                            0.1
-  steps:
-    - group:
-        name:                         Test preparation
-        steps:
-          - console:
-              name:                   Open the EUT console
-              console_name:           cons_target
-              doc:                    Initiates the console of the target in order
-                                      to be ready to capture its traces.
-              stop_on_failure:        True
-              steps:
-                - open:
-                    protocol: terminal
-
-    - loop:
-        name:                         Qualification loop
-        stop_on_failure:              False
-        steps:
-          - console:
-              name:                   write random data
-              console_name:           cons_target
-              steps:
-                  - writeln: bash $(test_directory)/generate_char.sh
-
-          - py_func:
-              name:                   Capture the terminal output
-              file:                   $(test_directory)/terminal_bug_reproducer.py
-              func_name:              RetreiveData
-              param:
-                - cons_target
-
-          - sleep: {timeout: 1}
-
-# Cleanup sequence
-#-------------------------------------------------------------------------------
-    - group:
-        name:                         Cleanup
-        execute_on_stop:              True
-        steps:
-          - console:
-              name:                   Close the target console
-              console_name:           cons_target
-              execute_on_stop:        True
-              steps:
-                - close:

test/validation/README.md

@@ -1,10 +1,76 @@
 # Validation
 
-This directory contains the necessary material to run the testium validation.
+This directory contains the testium validation suite. A single set of
+items (`items/`), fixtures and post-processing (`post_execution.py`) is
+re-used across every packaging channel.
 
-Here is the documentation on how to configure the validation, run it and check that the
-results are correct.
+## Running the suite
 
-# Tests
+```sh
+./test/validation/run.sh                       # default mode = source
+./test/validation/run.sh --mode wheel
+./test/validation/run.sh --mode pyinstaller
+./test/validation/run.sh --mode flatpak
+./test/validation/run.sh --mode appimage
+```
 
-TBD
+On Windows (only `source`, `wheel`, `pyinstaller` are supported):
+
+```bat
+test\validation\run.bat --mode pyinstaller
+```
+
+Pass `clean` as the **first** argument to recreate the validation venv
+from scratch (useful after a system Python upgrade):
+
+```sh
+./test/validation/run.sh clean --mode flatpak
+```
+
+Any extra arguments after the mode flag are forwarded to testium.
+
+## Modes
+
+| Mode          | What it launches                                            | Prerequisite                                                     |
+|---------------|-------------------------------------------------------------|------------------------------------------------------------------|
+| `source`      | `python3 src/testium` via the project's `run.sh`            | none — works straight out of the repo                            |
+| `wheel`       | `python -m testium` inside a dedicated wheel venv           | `./build_all.sh` produced `dist/testium-<v>-py3-none-any.whl`    |
+| `pyinstaller` | `dist/testium-<v>` (frozen binary)                          | `./build_all.sh` produced the PyInstaller binary                 |
+| `flatpak`     | `flatpak run --command=testium org.testium.Testium`         | the Flatpak bundle is installed (`flatpak install --user dist/testium-<v>.flatpak`) |
+| `appimage`    | `dist/Testium-<v>-x86_64.AppImage`                          | `./build_all.sh` produced the AppImage                           |
+
+Each mode writes its results to a distinct report file
+(`validation-<mode>.sqlite` / `validation-<mode>-<item>.xml`), so you
+can run several modes in a row without clobbering previous reports.
+
+## How `python_bin` is pinned
+
+Every test-execution subprocess (inline `<| ... |>` evaluation,
+`py_func`, `cycle`, `post_execution`, …) is routed through a dedicated
+venv at `${TMPDIR:-/tmp}/testium-validation-venv`. The venv is created
+with `--system-site-packages` so existing system packages stay visible,
+then `junit-xml` is pip-installed for `post_execution.py`.
+
+This is a **host** venv. In every mode (including Flatpak) the
+test-execution subprocesses end up running on the host — directly for
+source/wheel/pyinstaller/appimage, and via `flatpak-spawn --host` for
+Flatpak — so the same venv works across modes. The wheel mode
+additionally creates a separate `testium-wheel-venv-<v>` to hold the
+installed wheel; that one is only used to launch testium itself.
+
+## What is checked
+
+The `venv` item under `items/venv/` asserts that the validation venv is
+actually being used:
+
+* `python_bin` is set in the global dict.
+* The eval subprocess (used for `<| ... |>` expressions) has
+  `sys.executable == python_bin`, `sys.prefix == dirname(dirname(python_bin))`,
+  and `sys.prefix != sys.base_prefix` (i.e. is actually inside a venv).
+* A `py_func` subprocess passes the same three checks.
+
+These checks use `abspath`/`normpath` rather than `realpath` on
+purpose: the venv's `bin/python3` is a symlink to the host interpreter,
+so `realpath` would map both venv and non-venv interpreters to the same
+target. `sys.prefix != sys.base_prefix` is the venv-specific marker
+that distinguishes the two cases.

test/validation/items/console/test.tum

@@ -94,6 +94,17 @@
 {% endif %}
         - read_until: {expected: endOfCmd, timeout: 1, process_result: "'Hello' in r'''$(result)''' and 'PASS' in r'''$(result)''' "}
 
+{% if os == "Linux" %}
+- console:
+    name: Console runs on host (not the Flatpak sandbox)
+    doc: Regression guard for the 0.2.1 Flatpak bug (term console spawned inside the sandbox instead of on the host). /.flatpak-info exists only inside the sandbox, so the host-only marker is emitted (and matched by read_until) ONLY when the shell really runs on the host. On a broken Flatpak the marker never appears, read_until times out and the item FAILS. The marker is built at runtime ($M) so it is never present in the command line itself. Passes on every other channel.
+    console_name: term
+    key: $(test)_PASS
+    steps:
+        - writeln: 'test -e /.flatpak-info && M=SANDBOX || M=HOST; echo "console_host_check_$M"'
+        - read_until: {expected: console_host_check_HOST, timeout: 5}
+{% endif %}
+
 - console:
     name: Console closure
     execute_on_stop: true

test/validation/items/lua_func/lua_func.lua

@@ -49,4 +49,12 @@ function module.test_delgd()
     return 0
 end
 
+function module.return_nothing()
+    -- Returns no value: ret is nil but no error.
+end
+
+function module.return_explicit_nil()
+    return nil
+end
+
 return module

test/validation/items/lua_func/test.tum

@@ -186,6 +186,18 @@
     file: $(test_path)$(psep)lua_func.lua
     func_name: test_delgd
 
+- lua_func:
+    name: function returning nothing should succeed
+    key: $(test)_PASS
+    file: $(test_path)$(psep)lua_func.lua
+    func_name: return_nothing
+
+- lua_func:
+    name: function returning explicit nil should succeed
+    key: $(test)_PASS
+    file: $(test_path)$(psep)lua_func.lua
+    func_name: return_explicit_nil
+
 - group:
     name: context_id tests
     steps:

test/validation/items/py_func/py_func.py

@@ -54,3 +54,10 @@ def test_delgd():
     tm.delgd("_py_delgd_test")
     assert tm.gd("_py_delgd_test", None) is None
     return 0
+
+def return_nothing():
+    # Falls off the end: implicit None return, no error.
+    pass
+
+def return_explicit_none():
+    return None

test/validation/items/py_func/test.tum

@@ -196,6 +196,18 @@
     file: $(test_path)$(psep)py_func.py
     func_name: test_delgd
 
+- py_func:
+    name: function returning nothing should succeed
+    key: $(test)_PASS
+    file: $(test_path)$(psep)py_func.py
+    func_name: return_nothing
+
+- py_func:
+    name: function returning explicit None should succeed
+    key: $(test)_PASS
+    file: $(test_path)$(psep)py_func.py
+    func_name: return_explicit_none
+
 - group:
     name: context_id tests
     steps:

test/validation/items/venv/param.yaml

@@ -0,0 +1 @@
+no_param: Null

test/validation/items/venv/test.tum

@@ -0,0 +1,53 @@
+# venv test: assert that the dedicated validation venv is the python
+# being used for every test-execution subprocess (eval_proc / py_func /
+# cycle / ...). The venv path is pinned by ``-d python_bin=...`` in
+# test/validation/run.sh (or run.bat).
+#
+# We use ``abspath``/``normpath`` rather than ``realpath`` on purpose:
+# the venv's ``bin/python3`` is a symlink to the host python, so
+# realpath would map every venv interpreter to the same system path and
+# the comparison would silently pass even *without* a venv.
+# ``sys.prefix != sys.base_prefix`` is the venv-specific marker that
+# catches that case.
+
+- check:
+    name: python_bin is set in the global dict
+    key: $(test)_PASS
+    values:
+        - <| bool(r"$(python_bin)") |>
+
+- check:
+    name: eval_proc subprocess runs in the validation venv
+    key: $(test)_PASS
+    values:
+        - <| os.path.normpath(os.path.abspath(sys.executable)) == os.path.normpath(os.path.abspath(r"$(python_bin)")) |>
+
+- check:
+    name: eval_proc sys.prefix matches python_bin venv root
+    key: $(test)_PASS
+    values:
+        - <| os.path.normpath(os.path.abspath(sys.prefix)) == os.path.dirname(os.path.dirname(os.path.normpath(os.path.abspath(r"$(python_bin)")))) |>
+
+- check:
+    name: eval_proc is actually inside a venv (sys.prefix != sys.base_prefix)
+    key: $(test)_PASS
+    values:
+        - <| os.path.normpath(os.path.abspath(sys.prefix)) != os.path.normpath(os.path.abspath(sys.base_prefix)) |>
+
+- py_func:
+    name: py_func subprocess runs in the validation venv
+    key: $(test)_PASS
+    file: $(test_path)$(psep)verify_venv.py
+    func_name: check_sys_executable
+
+- py_func:
+    name: py_func sys.prefix matches python_bin venv root
+    key: $(test)_PASS
+    file: $(test_path)$(psep)verify_venv.py
+    func_name: check_sys_prefix_in_venv
+
+- py_func:
+    name: py_func is actually inside a venv
+    key: $(test)_PASS
+    file: $(test_path)$(psep)verify_venv.py
+    func_name: check_is_venv

test/validation/items/venv/verify_venv.py

@@ -0,0 +1,62 @@
+import os
+import sys
+
+import py_func.tm as tm
+
+
+def _norm(p):
+    # normpath + normcase, without resolving symlinks. realpath() would
+    # follow the venv's ``python3`` symlink to ``/usr/bin/python3.X`` and
+    # defeat the comparison.
+    return os.path.normcase(os.path.normpath(os.path.abspath(p)))
+
+
+def _venv_dir():
+    # python_bin is at ``<venv>/(bin|Scripts)/python*`` so the venv root
+    # is two levels above the executable.
+    exe = tm.gd("python_bin", "")
+    if not exe:
+        return ""
+    return os.path.dirname(os.path.dirname(_norm(exe)))
+
+
+def check_sys_executable():
+    """py_func subprocess: sys.executable must match the configured python_bin."""
+    expected = _norm(tm.gd("python_bin", ""))
+    actual = _norm(sys.executable)
+    if expected and actual == expected:
+        return True
+    return (
+        -1,
+        f"sys.executable={actual!r} differs from python_bin={expected!r}",
+    )
+
+
+def check_sys_prefix_in_venv():
+    """py_func subprocess: sys.prefix must match the venv root derived
+    from python_bin (two levels up from the executable)."""
+    venv = _venv_dir()
+    if not venv:
+        return (-1, "python_bin is not set in the global dict")
+    actual = _norm(sys.prefix)
+    if actual == venv:
+        return True
+    return (
+        -1,
+        f"sys.prefix={actual!r} is not the validation venv {venv!r}",
+    )
+
+
+def check_is_venv():
+    """py_func subprocess: confirm we are inside a venv, i.e. sys.prefix
+    differs from sys.base_prefix. This catches the case where python_bin
+    happens to be a system interpreter and the path-based check would
+    pass trivially."""
+    actual = _norm(sys.prefix)
+    base = _norm(sys.base_prefix)
+    if actual != base:
+        return True
+    return (
+        -1,
+        f"sys.prefix == sys.base_prefix == {actual!r}: not running in a venv",
+    )

test/validation/lsp_check.py

@@ -0,0 +1,102 @@
+#!/usr/bin/env python3
+"""Per-channel check of the testium language server.
+
+Given the channel's testium invocation as argv (e.g. ``flatpak run
+--command=testium org.testium.Testium``, a PyInstaller binary path, or
+``python -m testium``), verify two things end-to-end against that exact build:
+
+  1. ``<cmd> schema`` produces valid JSON whose item registry still includes the
+     nested action sets (console/plot/json_rpc). This catches a frozen build
+     that lost the actions — the failure mode the declarative ``ACTIONS``
+     refactor fixed (no more ``inspect.getsource`` at runtime).
+  2. ``<cmd> lsp`` starts a real language server: it must answer an LSP
+     ``initialize`` request with a capabilities result and must NOT report the
+     pygls dependency as missing. This catches a channel that forgot to bundle
+     the ``[lsp]`` extra.
+
+Exits non-zero (with a diagnostic) on the first failure so the validation run
+fails loudly. Used by ``run.sh`` before launching the main suite.
+"""
+import json
+import subprocess
+import sys
+
+EXPECTED_ACTION_PARENTS = ("console", "plot", "json_rpc")
+
+
+def fail(msg):
+    print(f"LSP CHECK: FAIL — {msg}", file=sys.stderr)
+    sys.exit(1)
+
+
+def _extract_json(raw):
+    """Parse JSON from ``raw`` bytes, tolerating leading non-JSON noise.
+
+    The source-mode launcher (run.sh) may print env-setup lines before the
+    schema JSON, so we fall back to parsing from the first ``{``.
+    """
+    try:
+        return json.loads(raw)
+    except json.JSONDecodeError:
+        start = raw.find(b"{")
+        if start < 0:
+            raise
+        return json.loads(raw[start:])
+
+
+def check_schema(cmd):
+    try:
+        out = subprocess.run(cmd + ["schema"], capture_output=True, timeout=120)
+    except Exception as e:  # noqa: BLE001
+        fail(f"`{' '.join(cmd)} schema` could not run: {e}")
+    if out.returncode != 0:
+        fail(f"`schema` exited {out.returncode}: {out.stderr.decode()[:300]}")
+    try:
+        data = _extract_json(out.stdout)
+    except json.JSONDecodeError as e:
+        fail(f"`schema` output is not valid JSON: {e}")
+    items = data.get("items", {})
+    for parent in EXPECTED_ACTION_PARENTS:
+        actions = (items.get(parent) or {}).get("actions") or {}
+        if not actions:
+            fail(f"schema item '{parent}' has no actions — a frozen build lost "
+                 f"the declarative ACTIONS (item keys: {sorted(items)[:8]}…)")
+    print(f"LSP CHECK: schema OK ({len(items)} items; actions present for "
+          f"{', '.join(EXPECTED_ACTION_PARENTS)})")
+
+
+def check_lsp(cmd):
+    body = json.dumps({
+        "jsonrpc": "2.0", "id": 1, "method": "initialize",
+        "params": {"processId": None, "rootUri": None, "capabilities": {}},
+    }).encode()
+    msg = b"Content-Length: %d\r\n\r\n%s" % (len(body), body)
+    try:
+        out = subprocess.run(cmd + ["lsp"], input=msg,
+                             capture_output=True, timeout=30)
+        stdout, stderr = out.stdout, out.stderr
+    except subprocess.TimeoutExpired as e:
+        # A server that stays alive past initialize is fine — it just never saw
+        # a shutdown. Use whatever it wrote so far as the response.
+        stdout, stderr = e.stdout or b"", e.stderr or b""
+    blob = stdout + stderr
+    if b"dependencies missing" in blob:
+        fail("`lsp` reports the pygls dependency missing — this channel did "
+             "not bundle the [lsp] extra.")
+    if b'"capabilities"' not in stdout:
+        fail("`lsp` did not return an initialize result. "
+             f"stdout[:200]={stdout[:200]!r} stderr[:200]={stderr[:200]!r}")
+    print("LSP CHECK: lsp initialize OK (server answered with capabilities)")
+
+
+def main():
+    cmd = sys.argv[1:]
+    if not cmd:
+        fail("usage: lsp_check.py <testium-invocation...>")
+    check_schema(cmd)
+    check_lsp(cmd)
+    print("LSP CHECK: PASS")
+
+
+if __name__ == "__main__":
+    main()

test/validation/run.bat

@@ -0,0 +1,131 @@
+@echo off
+SETLOCAL EnableExtensions EnableDelayedExpansion
+
+REM Runs the testium validation suite against any installable channel of
+REM testium on Windows (source, wheel, pyinstaller).
+REM
+REM Usage:
+REM   test\validation\run.bat [clean] [--mode MODE] [extra testium args]
+REM
+REM   clean       remove the validation venv before recreating it
+REM               (must be the first argument; useful after a Python upgrade)
+REM
+REM   --mode MODE which testium build to validate. One of:
+REM                   source       (default) project's run.bat (src\testium)
+REM                   wheel        dist\testium-<v>-py3-none-any.whl
+REM                   pyinstaller  dist\testium-<v>.exe (or dist\testium-<v>)
+REM
+REM Every test-execution subprocess runs in a dedicated host venv under
+REM %TEMP%\testium-validation-venv (created with --system-site-packages,
+REM then junit-xml is pip-installed for post_execution.py).
+REM
+REM The report file is suffixed with the mode so consecutive runs in
+REM different modes don't overwrite each other.
+
+SET "SCRIPT_DIR=%~dp0"
+IF "%SCRIPT_DIR:~-1%"=="\" SET "SCRIPT_DIR=%SCRIPT_DIR:~0,-1%"
+SET "PROJECT_DIR=%SCRIPT_DIR%\..\.."
+SET /P VERSION=<"%PROJECT_DIR%\src\VERSION"
+
+REM ---------- arg parsing ----------------------------------------------------
+
+SET "MODE=source"
+SET "CLEAN=0"
+IF /I "%~1"=="clean" (
+    SET "CLEAN=1"
+    SHIFT
+)
+
+SET "EXTRA="
+:PARSE_ARGS
+IF "%~1"=="" GOTO ARGS_DONE
+IF /I "%~1"=="--mode" (
+    SET "MODE=%~2"
+    SHIFT
+    SHIFT
+    GOTO PARSE_ARGS
+)
+SET "EXTRA=!EXTRA! "%~1""
+SHIFT
+GOTO PARSE_ARGS
+:ARGS_DONE
+
+REM ---------- locate host python ---------------------------------------------
+
+SET "PYTHON_EXE="
+py --version >nul 2>&1
+IF %ERRORLEVEL% EQU 0 (
+    SET "PYTHON_EXE=py"
+    GOTO PYTHON_FOUND
+)
+python --version >nul 2>&1
+IF %ERRORLEVEL% EQU 0 (
+    SET "PYTHON_EXE=python"
+    GOTO PYTHON_FOUND
+)
+echo ERROR: Python could not be found on this system.
+exit /b 1
+:PYTHON_FOUND
+
+REM ---------- validation venv -------------------------------------------------
+
+SET "VENV_DIR=%TEMP%\testium-validation-venv"
+IF "%CLEAN%"=="1" IF EXIST "%VENV_DIR%" rmdir /s /q "%VENV_DIR%"
+
+IF NOT EXIST "%VENV_DIR%" (
+    echo Creating validation venv at %VENV_DIR%
+    %PYTHON_EXE% -m venv --system-site-packages "%VENV_DIR%"
+    IF !ERRORLEVEL! NEQ 0 (
+        echo ERROR while creating the validation venv.
+        exit /b 1
+    )
+    call "%VENV_DIR%\Scripts\pip" install --quiet --upgrade pip
+    call "%VENV_DIR%\Scripts\pip" install --quiet junit-xml
+)
+SET "VENV_PYTHON=%VENV_DIR%\Scripts\python.exe"
+
+REM ---------- shared "tail" forwarded to every launcher -----------------------
+REM Reports are stamped with the mode so successive runs don't clobber each other.
+
+SET "TAIL=-b -d "python_bin=%VENV_PYTHON%" -d "validation_report_file=validation-%MODE%" -- "%SCRIPT_DIR%\main.tum"%EXTRA%"
+
+REM ---------- per-mode launcher ----------------------------------------------
+
+echo -- validation mode: %MODE%
+
+IF /I "%MODE%"=="source"      GOTO MODE_SOURCE
+IF /I "%MODE%"=="wheel"       GOTO MODE_WHEEL
+IF /I "%MODE%"=="pyinstaller" GOTO MODE_PYI
+echo ERROR: unknown --mode '%MODE%'. Expected: source ^| wheel ^| pyinstaller.
+exit /b 1
+
+:MODE_SOURCE
+call "%PROJECT_DIR%\run.bat" %TAIL%
+exit /b %ERRORLEVEL%
+
+:MODE_WHEEL
+SET "WHEEL=%PROJECT_DIR%\dist\testium-%VERSION%-py3-none-any.whl"
+IF NOT EXIST "%WHEEL%" (
+    echo ERROR: wheel not found at %WHEEL% -- run build_all.sh first.
+    exit /b 1
+)
+SET "WHEEL_VENV=%TEMP%\testium-wheel-venv-%VERSION%"
+IF "%CLEAN%"=="1" IF EXIST "%WHEEL_VENV%" rmdir /s /q "%WHEEL_VENV%"
+IF NOT EXIST "%WHEEL_VENV%" (
+    echo Creating wheel venv at %WHEEL_VENV%
+    %PYTHON_EXE% -m venv --system-site-packages "%WHEEL_VENV%"
+    call "%WHEEL_VENV%\Scripts\pip" install --quiet --upgrade pip
+    call "%WHEEL_VENV%\Scripts\pip" install --quiet "%WHEEL%"
+)
+"%WHEEL_VENV%\Scripts\python.exe" -m testium %TAIL%
+exit /b %ERRORLEVEL%
+
+:MODE_PYI
+SET "PYI_BIN=%PROJECT_DIR%\dist\testium-%VERSION%.exe"
+IF NOT EXIST "%PYI_BIN%" SET "PYI_BIN=%PROJECT_DIR%\dist\testium-%VERSION%"
+IF NOT EXIST "%PYI_BIN%" (
+    echo ERROR: PyInstaller binary not found in %PROJECT_DIR%\dist -- run build_all.sh first.
+    exit /b 1
+)
+"%PYI_BIN%" %TAIL%
+exit /b %ERRORLEVEL%

test/validation/run.sh

@@ -0,0 +1,153 @@
+#!/bin/bash
+# Runs the testium validation suite against any installable channel of
+# testium (source, wheel, pyinstaller, flatpak, appimage).
+#
+# Usage:
+#   ./test/validation/run.sh [clean] [--mode MODE] [extra testium args]
+#
+#   clean           remove the validation venv before recreating it
+#                   (must be the first argument; useful after a Python upgrade)
+#
+#   --mode MODE     which testium build to validate. One of:
+#                       source       (default) src/testium via project run.sh
+#                       wheel        dist/testium-<v>-py3-none-any.whl
+#                       pyinstaller  dist/testium-<v>
+#                       flatpak      installed org.testium.Testium
+#                       appimage     dist/Testium-<v>-*.AppImage
+#
+# Every test-execution subprocess (inline <| ... |>, py_func, cycle,
+# post_execution, ...) runs in a dedicated host venv under
+# /tmp/testium-validation-venv. That venv is shared across modes —
+# even Flatpak reaches it via flatpak-spawn --host. The validation venv
+# is created with --system-site-packages so existing system packages
+# (PySide6, lxml, ...) stay visible, then junit-xml is pip-installed
+# for post_execution.py.
+#
+# The report file is suffixed with the mode (e.g. validation-flatpak.sqlite)
+# so consecutive runs in different modes don't overwrite each other.
+
+set -e
+
+SCRIPT_PATH="$(readlink -f "$0")"
+SCRIPT_DIR="$(realpath "$(dirname "$SCRIPT_PATH")")"
+PROJECT_DIR="$(realpath "$SCRIPT_DIR/../..")"
+VERSION="$(cat "$PROJECT_DIR/src/VERSION")"
+
+# ---------- arg parsing -------------------------------------------------------
+
+MODE=source
+
+if [ "${1:-}" = "clean" ]; then
+    CLEAN=1
+    shift
+else
+    CLEAN=0
+fi
+
+EXTRA=()
+while [ $# -gt 0 ]; do
+    case "$1" in
+        --mode)
+            MODE="$2"
+            shift 2
+            ;;
+        --mode=*)
+            MODE="${1#--mode=}"
+            shift
+            ;;
+        *)
+            EXTRA+=("$1")
+            shift
+            ;;
+    esac
+done
+
+# ---------- validation venv ---------------------------------------------------
+
+VENV_DIR="${TMPDIR:-/tmp}/testium-validation-venv"
+if [ "$CLEAN" -eq 1 ]; then
+    rm -rf "$VENV_DIR"
+fi
+
+if [ ! -d "$VENV_DIR" ]; then
+    echo "Creating validation venv at $VENV_DIR"
+    python3 -m venv --system-site-packages "$VENV_DIR"
+    "$VENV_DIR/bin/pip" install --quiet --upgrade pip
+    "$VENV_DIR/bin/pip" install --quiet junit-xml
+fi
+VENV_PYTHON="$VENV_DIR/bin/python3"
+
+# ---------- per-mode launcher -------------------------------------------------
+
+case "$MODE" in
+    source)
+        CMD=("$PROJECT_DIR/run.sh")
+        ;;
+    wheel)
+        WHEEL="$PROJECT_DIR/dist/testium-${VERSION}-py3-none-any.whl"
+        if [ ! -f "$WHEEL" ]; then
+            echo "ERROR: wheel not found at $WHEEL — run ./build_all.sh first." >&2
+            exit 1
+        fi
+        WHEEL_VENV="${TMPDIR:-/tmp}/testium-wheel-venv-${VERSION}"
+        if [ "$CLEAN" -eq 1 ]; then
+            rm -rf "$WHEEL_VENV"
+        fi
+        if [ ! -d "$WHEEL_VENV" ]; then
+            echo "Creating wheel venv at $WHEEL_VENV"
+            python3 -m venv --system-site-packages "$WHEEL_VENV"
+            "$WHEEL_VENV/bin/pip" install --quiet --upgrade pip
+            # Install with the [lsp] extra so the wheel channel is validated in
+            # its language-server-capable form (pulls pygls), matching how a
+            # user enables `testium lsp` from a wheel: pip install testium[lsp].
+            "$WHEEL_VENV/bin/pip" install --quiet "${WHEEL}[lsp]"
+        fi
+        CMD=("$WHEEL_VENV/bin/python" -m testium)
+        ;;
+    pyinstaller)
+        PYI_BIN="$PROJECT_DIR/dist/testium-${VERSION}"
+        if [ ! -x "$PYI_BIN" ]; then
+            echo "ERROR: PyInstaller binary not found at $PYI_BIN — run ./build_all.sh first." >&2
+            exit 1
+        fi
+        CMD=("$PYI_BIN")
+        ;;
+    flatpak)
+        if ! flatpak info --user org.testium.Testium &>/dev/null \
+           && ! flatpak info --system org.testium.Testium &>/dev/null; then
+            echo "ERROR: org.testium.Testium is not installed." >&2
+            echo "       flatpak install --user $PROJECT_DIR/dist/testium-${VERSION}.flatpak" >&2
+            exit 1
+        fi
+        CMD=(flatpak run --command=testium org.testium.Testium)
+        ;;
+    appimage)
+        APPIMAGE=$(ls -1t "$PROJECT_DIR/dist"/Testium-"${VERSION}"-*.AppImage 2>/dev/null | head -1)
+        if [ -z "$APPIMAGE" ] || [ ! -x "$APPIMAGE" ]; then
+            echo "ERROR: no AppImage for version $VERSION under $PROJECT_DIR/dist — run ./build_all.sh first." >&2
+            exit 1
+        fi
+        CMD=("$APPIMAGE")
+        ;;
+    *)
+        echo "ERROR: unknown --mode '$MODE'. Expected: source|wheel|pyinstaller|flatpak|appimage." >&2
+        exit 1
+        ;;
+esac
+
+# ---------- launch ------------------------------------------------------------
+
+echo "-- validation mode: $MODE"
+echo "-- launch: ${CMD[*]}"
+
+# ---------- LSP check (this exact channel) ------------------------------------
+# Verify `testium lsp` / `testium schema` work in the build under test before
+# running the suite: schema must keep its nested actions (declarative ACTIONS,
+# survives frozen builds) and the language server must start (pygls bundled).
+echo "-- LSP check ($MODE)"
+"$VENV_PYTHON" "$SCRIPT_DIR/lsp_check.py" "${CMD[@]}"
+
+exec "${CMD[@]}" -b \
+    -d "python_bin=$VENV_PYTHON" \
+    -d "validation_report_file=validation-$MODE" \
+    -- "$SCRIPT_DIR/main.tum" "${EXTRA[@]}"