Translation sync failed for PR #821 (zh-cn)

[mmcky]

Translation sync failure

Source PR: #821
Target repo: QuantEcon/lecture-intro.zh-cn
Target language: zh-cn

Errors

• Error processing lectures/prob_dist.md: Error: structural parity check failed for lectures/prob_dist.md:
  • directive count differs: source has 55, output has 36 — a wholesale mismatch usually means the document was re-fenced or truncated
  • directive Add production network lecture with IO visualizations #6 name changed: source line 243 has {note}, output line 143 has {code-cell}
  • directive Add a lecture on inequality measures #14 name changed: source line 366 has {code-cell}, output line 324 has {solution-start}
  • directive Add lecture on inequality measures #16 name changed: source line 379 has {code-cell}, output line 344 has {solution-end}
  • directive Add a simplified lecture on Markov chains #23 name changed: source line 461 has {exercise}, output line 484 has {code-cell}
  • directive Add a lecture on eigenvals, eigenvectors, Perron-Frobenius and the Neumann series theorem #24 name changed: source line 467 has {solution-start}, output line 489 has {exercise}
  • directive Add a simple lecture introducing networks #26 name changed: source line 487 has {solution-end}, output line 527 has {code-cell}

Recovery

Once the issue is resolved, comment \translate-resync on PR #821 to retry.

[mmcky]

Route corrected — and this lecture needs more than a retry

Per the 2026-08-20 ruling on action-translation#276, \translate-resync is no longer the documented recovery for a failed sync, and it would not have cleared this one: structural-parity failures reproduce byte-for-byte on retry.

Status — worse than the state file suggests. prob_dist.md on lecture-intro.zh-cn records source-sha: 6c607ded (the 2026-07-18 bootstrap) against a source last-touch of ebff90fd. Four source commits are unlanded: fd3241e (#811, the Observed Distributions split, −287/+157 lines), c8587ed (#816, figure captions, +98), a453cf5 (#819, sequence linking) and ebff90f (#821, the Japan data read).

Only the last of those reached the target, by hand in lecture-intro.zh-cn#298. Measured on origin/main: source prob_dist.md carries 19 caption: values against the target's 3, and the target is 804 lines against the source's 1020.

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

Sequencing caution. Recovering this lecture delivers ~16 new figure captions, which walks straight into the open action-translation#255 class — captions arriving byte-identical to English. The three captions the target already has are correctly localised, so the risk is confined to the new ones. Worth doing after the shared caption check lands, or with an explicit caption pass on the result.

This issue is one of four filed against the same lecture (#812, #818, #820, #822) as successive source PRs each failed on it; they share one recovery.

[mmcky]

Recovered and closing. prob_dist.md is resynced to source ebff90fd on lecture-intro.zh-cn@main as of lecture-intro.zh-cn#300 (merged 2026-08-20).

This issue is one of four filed against the same lecture as successive source PRs each failed on it — #812, #818, #820, #822. One recovery clears all four; they are being closed together.

Delivered with translate forward -f at engine v0.26.0, not \translate-resync, per the ruling on action-translation#276.

The gap was larger than the state file suggested: three of the four source changes had never landed, so the target was 216 lines short and carried 3 figure captions against source's 19. Only the data-lectures reads from #821 had reached it, applied by hand in lecture-intro.zh-cn#298 — and those hand fixes survived the regeneration intact.

Verified: 55/55 directives, 49/49 code cells, 14/14 headings, 0 python syntax errors across all 49 cells, 19/19 figure captions localised with none byte-identical to source, and the i18n font block preserved.