Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if (isJassFile(filename)) { | ||
| Map<String, String> newFunctions = collectJassFunctions(cu); | ||
| Map<String, String> oldFunctions = jassFunctionSnapshots.getOrDefault(filename, Collections.emptyMap()); | ||
| Set<String> changedFunctions = determineChangedJassFunctions(oldFunctions, newFunctions); | ||
| pendingJassFunctionChanges.put(filename, changedFunctions); | ||
| jassFunctionSnapshots.put(filename, newFunctions); | ||
| } else { |
There was a problem hiding this comment.
Jass fan‑out ignores non‑function declarations
The change detection for Jass files only fingerprints FunctionDefinitions (collectJassFunctions / determineChangedJassFunctions) and calculateCUsToUpdate only propagates updates when one of those functions changes. If a Jass file modifies or removes other top‑level declarations—e.g. globals, native declarations, or struct typedefs—pendingJassFunctionChanges remains empty, so no dependent Wurst compilation units are rechecked. Previously any edit in a .j file forced a full rebuild, so these dependencies were safe. After this commit, packages that rely on a changed global can continue using stale attribute caches and miss the resulting errors until a manual rebuild, which is a regression in correctness. Consider hashing all relevant Jass declarations or falling back to the previous whole‑project invalidation when non‑function changes are detected.
Useful? React with 👍 / 👎.
No description provided.