Translation sync failed for PR #595 (fa) #614

Activity

1. mmcky

   added

   translation-sync-failure

   on Aug 18, 2026
2. mmcky commented on Aug 20, 2026

   

   mmcky

   on Aug 20, 2026

   ContributorAuthor

   More actions

   Route corrected — use the CLI, not \translate-resync

   Per the 2026-08-20 ruling on action-translation#276, \translate-resync is no longer the documented recovery, and it would not have cleared this: structural-parity failures reproduce byte-for-byte on retry. This issue and its sibling are that demonstration — one was filed by the original run at engine v0.25.0 and one by the resync retry at v0.26.0, and their bodies differ only by every output line number shifting by one.

   Status. about_py.md on lecture-python-programming.fa is behind: state records source-sha: 3213613a (2026-05-09) against a source last-touch of d35eb831 (PR #595).

   Route. translate forward -f about_py.md from a source clone pulled to origin/main — filesystem-sourced, immune to #276, parity anchored to current main.

   Caution. This is also the file that repeatedly fails the rebase automation with structural parity check failed for lectures/about_py.md, leaving sibling PRs un-rebasable. If forward hits the same parity violation, that is the real defect to file rather than something to work around — the violation list in this issue's body is the starting evidence.
3. mmcky commented on Aug 20, 2026

   

   mmcky

   on Aug 20, 2026

   ContributorAuthor

   More actions

   Recovered and closing — with no re-translation at all.

   about_py.md on lecture-python-programming.fa was never actually behind in content. Source PR #595 changed exactly two lines of this lecture, the capitalisation Pytorch → PyTorch, and that fix was already present in the edition, applied by hand in fa#124 on 2026-06-19. Only the state file was stale, at 3213613a. lecture-python-programming.fa#160 advances it to d35eb831 (merged 2026-08-20).

   A translate forward run was attempted first and deliberately discarded. It rewrote 266 lines to deliver a two-character casing fix that was already applied, and it dropped translator content — the tool's own triage flagged an added PyTorch/MATLAB sentence and a Java code comment before writing, and git log traces that content to Zahra's 2025-10-30 commit (fa#14). Per the ruling on action-translation#276, a natively-translated lecture is not regenerated when a deterministic repair will do.

   Worth noting: this issue and its sibling are the estate's clearest demonstration that a resync retry does not clear a structural parity failure. One was filed by the original run at engine v0.25.0 and one by the \translate-resync retry at v0.26.0, and their bodies differ only by every output line number shifting by one.

   This does not address the separate defect where about_py.md repeatedly fails the rebase automation with the same parity check, leaving sibling PRs un-rebasable.
4. mmcky

   closed this as completedon Aug 20, 2026