Update using-solid-node.rst#7
Open
felipesanches wants to merge 1 commit into
Open
Conversation
lfagundes
added a commit
that referenced
this pull request
Jul 16, 2026
…b viewer Bug (improvements.md #7): `solid develop` respawns a fresh Builder subprocess on every save. When re-importing edited project source raised at module level (NameError, SyntaxError, ImportError, ...), Builder._start() caught the exception and awaited self.file_changed -- but the watchdog Observer is only scheduled *after* a successful load_node()+assemble(), which is exactly what had just failed. So the future could never resolve: the builder subprocess hung forever with nothing watching, and Develop.handle()'s `builder_proc.join()` blocked forever too, silently freezing the whole develop loop (no further saves were ever picked up). Worse, on the very next loop iteration Develop.handle() unconditionally restarts the web viewer ("Restarting WEB"). WebViewer.__init__ calls load_node(path) itself with no error handling, so if the project was still broken at that moment the web viewer subprocess crashed outright -- taking the browser-facing server, the reload websocket and the /_build_error endpoint down with it. Fix: - Builder gets an `is_reload` flag. Develop.handle() passes False only for the very first attempt of a `solid develop` invocation and True for every attempt afterwards (every attempt the watch loop triggers). A load/assemble exception on a reload attempt now falls back to watching the whole project directory recursively (filtered to .py, ignoring __pycache__, since the precise file list is exactly what failed to be discovered), writes the error to the existing errors.json channel via report_error(), and exits cleanly the moment a fix is saved so Develop respawns and retries. On the very first (non-reload) attempt the old semantics are kept intentionally fragile, but no longer hang: a broken project logs one clean line and exits non-zero instead of dumping a traceback and hanging forever. - Develop.handle() now checks the first builder_proc's exitcode: if that initial attempt fails, it tears down whatever else it started (web/web-dev/openscad) and exits the whole command, instead of looping around and treating the next attempt as a recoverable reload. - WebViewer.__init__ now guards load_node()+NodeAPI construction: a broken project no longer crashes the webserver subprocess. It logs locally and simply skips mounting /node for that instance -- the existing /_build_error endpoint (fed by Builder's errors.json) and the reload websocket stay up, and the next successful reload restarts the viewer with a working node again. - Also guards Builder.on_modified's Future.set_result against a second filesystem event arriving before the first is consumed. Tests (TDD, verified failing against the pre-fix code by stashing the source changes and re-running): - tests/test_builder_reload_resilience.py: drives the real Builder (real watchdog Observer, real asyncio loop) against a scratch copy of tests/flat_project, each attempt as a real multiprocessing.Process (so sys.exit() only ends that child, matching production). Covers a NameError reload and a SyntaxError reload staying alive and surfacing the error, recovery clearing the error and building after a fix, and the startup (non-reload) case exiting promptly instead of hanging. - tests/test_manager_develop.py: unit tests that Develop.handle() flags only the first builder invocation as non-reload, and that a failed first attempt tears down the other processes and exits. - tests/test_web_viewer.py: WebViewer construction, /_build_error and the reload websocket all survive a load_node() failure. Gates: `.venv/bin/python -m pytest tests/ -q` -- 169 passed. Gearbox canary (all 8 node files under examples/gearbox/root) -- each ends `0 failed`. Manual verification: ran `solid develop` against a scratch copy of tests/flat_project, broke simple_pipe.py with a module-level NameError mid-edit, confirmed the process stayed alive and /_build_error surfaced the traceback while /node/ kept responding, fixed the file, and confirmed /_build_error cleared and /node/ + the STL route served the rebuilt model again -- server never dropped connection throughout. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.