Repository navigation
fix(live): avoid contextual transcript casing scans - #1195
Conversation
Castiron custom codeEvaluated main: Trusted main policy: ✅ No new custom-code files detected. 96 mixed files remain; 0 existing customizations changed. Compared 96 existing customizations unchanged
56 more in the full report. A changed generated baseline means this report cannot reliably identify which handwritten lines changed. Inspect the custom-code diffDownload the exact patch produced by this run (requires repository access): gh run download 38926856818 --repo openai/openai-java \
--name castiron-custom-code-38926856818-1 --dir /tmp/castiron-custom-code-38926856818-1
git apply --stat /tmp/castiron-custom-code-38926856818-1/custom-code.patch
cat /tmp/castiron-custom-code-38926856818-1/custom-code.patchOr reproduce it from an SDK checkout containing the vendored reporter: git fetch --no-tags origin 848ef53c94498e795587d571900202ea383601d4 4b0829fe433f37672d944db44a56407bfe48d481
python3 scripts/castiron/custom_code_report.py report \
--base 848ef53c94498e795587d571900202ea383601d4 \
--head 4b0829fe433f37672d944db44a56407bfe48d481 --fetch --require-head-hash --public \
--out /tmp/castiron-custom-code-4b0829fe433f
cat /tmp/castiron-custom-code-4b0829fe433f/custom-code.patchThis is the current full custom patch for mixed files, not an attribution of only the handwritten lines changed by this PR. |
Castiron-Internal-PR: openai/openai-java-internal#281 Castiron-Source-SHA: e8eed1e3442f6a92193179487d796912242cae6d Castiron-Public-Base-SHA: 848ef53
1f96fb0 to
4b0829f
Compare
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
jbeckwith-oai
left a comment
jbeckwith-oai
left a comment
There was a problem hiding this comment.
Reviewed all three changed files at 4b0829f; no actionable findings.
Pre-folding uppercase Sigma preserves the existing sigma-equivalence rule while avoiding the contextual lowercase path in OpenJDK. Configured phrases and streamed fragments use the same normalization; dotted-I expansion, split supplementary characters, and original emitted transcript text remain preserved. The new long-Sigma regression complements the existing fragmented-Unicode cases. The old and new public generation snapshots have identical trees.
Validation: static source/test inspection and clean diff check. Exact-head hosted tests, runtime/API/Jackson compatibility, lint/build and Castiron checks passed. I did not execute the candidate or reproduce the timing assertion locally.
markstuart-oai
left a comment
markstuart-oai
left a comment
There was a problem hiding this comment.
Reviewed 4b0829fe433f37672d944db44a56407bfe48d481. Folding capital sigma before lowercase conversion avoids the contextual scan while preserving the existing sigma matching rules. Both normalization paths use the same helper, and the transcript output keeps its original text.
No actionable findings in the full three-file diff and surrounding code. The regression covers a long sigma transcript; existing Unicode and fragment tests cover matching behavior. Hosted checks passed in the captured run. This was a source review; I did not execute candidate code.
3 participants
Avoid contextual case-folding scans when matching Live transcript acknowledgments. Preserve Greek sigma equivalence, fragmented Unicode handling, and original transcript text.