chore: sync with upstream pingdotgg/t3code main - #53
Conversation
yordis
commented
Sep 18, 2026
- Upstream now reads viewer access once, off the pull request detail it already fetches, so the fork's separate permission read has no caller left. The fork's administrator bit is what the merge-bypass affordance actually depends on, so it moves onto the shape upstream kept rather than being carried alongside a read nothing calls.
- Two ledger entries are gone because upstream shipped their behavior: server log records reaching a collector, and a browser trace staying attached to the server trace that started it. Keeping an entry for something upstream now owns would claim a divergence that no longer exists.
- The part of the logs entry upstream did not take, per-signal configuration for log records and their own batching knobs, folds into the entry about honoring the standard OpenTelemetry variables, which is where the rest of that surface is already described.
- The local file tracer no longer also receives log records once OTLP log export is on. Upstream treats the double export as the documented tradeoff, and disagreeing would mean maintaining a second answer to a question upstream has already answered.
- The lockfile is regenerated rather than hand merged. The fork carries a dependency patch upstream does not, and its integrity hash is derived from the patch contents, so only a real install recomputes it correctly.
Co-authored-by: Julius Marminge <51714798+juliusmarminge@users.noreply.github.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
…ngdotgg#12181) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Julius Marminge <51714798+juliusmarminge@users.noreply.github.com>
…pingdotgg#12333) Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
…requests (pingdotgg#11050) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
…#11842) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
…2315) Co-authored-by: Cursor <cursoragent@cursor.com>
…11825) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…ingdotgg#11665) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Julius Marminge <51714798+juliusmarminge@users.noreply.github.com>
Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com>
Upstream now reads viewer access once off the pull request detail, so the fork's extra permission read has no caller left and its administrator bit belongs on the shape upstream kept. Upstream landed its own server log export and client trace propagation, so those two ledger entries describe upstream behavior rather than a divergence. The lockfile is regenerated because the fork's only patch upstream does not carry needs its hash recomputed from the patch itself, which a hand merge cannot do. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com>
PR SummaryLow Risk Overview Review and vendored reference noise is reduced: CodeRabbit and new CI/ops tweaks: relay production deploy uses Reviewed by Cursor Bugbot for commit 08065b7. Bugbot is set up for automated code reviews on this repo. Configure here. |