mirror of
https://github.com/logos-co/logos-plugin-qt.git
synced 2026-08-27 17:01:11 +00:00
Dario Gabriel LipicarandClaude Opus 5
f668ef212f
#28 fixed one of two. The single-concurrency path returns early -- `if (kVoidMethods.contains(methodName)) return QVariant(true);` -- before reaching the branch that patch changed, and its own comment said the quiet part out loud: "void methods answer QVariant(true) WHATEVER the cdylib put on the C ABI". That is the one case where the cdylib has something to say. Measured, which is the only reason this was caught: with #28 merged, on plugin-qt master, reached through module-builder b06c287 and pinned by logos-test-modules -- verified by following the lock root-to-leaf, and by content (`__rejected` present in the emitter) -- the six failure/B/arity/too-many-zero-parameter cells were STILL xfail. The fix was in the build and the cells had not moved, so the fix was not on their path. THE TEST IS WHY IT LOOKED DONE, twice over, and both are fixed here: * It asserted `grep -q '__rejected'`, which passes on one site as happily as on two. * The first attempt to strengthen that counted occurrences in $cm -- the MULTI glue file -- while the site #28 missed emits into $c, the SINGLE one. Checking the right property in the wrong file is the same mistake one step over, and the suite caught it. Each site is now asserted against the file it lives in, so patching one path again fails here rather than at the far end of a matrix run. An ordinary void reply still answers QVariant(true) on both paths; only a provider refusal -- the closed set dispatch_failed / invalid_args / unknown_method, matched on an object with a string `code` -- is propagated. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>