8 lines
1.7 KiB
Plaintext
8 lines
1.7 KiB
Plaintext
Initial rebuild rejected an out-of-range JoinImageWithAlpha citation (226, then 225 vs actual223). Corrected to188-223 before accepting definition. No source/graph was modified.
|
|
Initial backend run: 281 passed, 2 failed. Both were test defects: text searched 未显式result while flow said 无显式result; some old definitions have no comments key, so frozen notes are counted via actual build_document/source_notes. Corrected without relaxing preserved old review equality or failure closure.
|
|
Initial Ruff rejected dense source formatting, unused ref assignments and long literals. During cleanup a substring replacement temporarily changed comparereg into compareref; corrected before final test/build. All upstream text and diagnostic logs kept raw, authored Python reformatted. No skips or rule suppression added.
|
|
First full browser pass: 1334 assertions. Then the default overview was rewritten in plain language, the seed control schema default corrected to accurate entry fallback randomize (source fixed remains distinct; before/after depends on setting), citations narrowed and PreviewImage/rgthree registration references completed. Final test/build/protection/browser results refer to that final data.
|
|
No full backend/frontend/conversion run was necessary: execution/conversion/Drawer code are frozen. Historical 5 business failures and older I001 findings are inherited evidence, not a new run or a claim that whole repo is green.
|
|
|
|
Final manual schema reread found ResizeImageMaskNode.scale_method has explicit default area (saved lanczos). Corrected the previously unspecified-default text, added saved-vs-default semantic assertion, and reran final definition/lint/test/build/protection/repeat/browser validation. No source saved value changed.
|