Skip to content

Translation sync failed for PR #1006 (zh-cn) #1007

Closed

Description

@mmcky

Translation sync failure

Source PR: #1006
Target repo: QuantEcon/lecture-python.zh-cn
Target language: zh-cn

Errors

Recovery

Once the issue is resolved, comment \translate-resync on PR #1006 to retry.

Activity

  1. mmcky commented on Aug 20, 2026

    @mmcky
    ContributorAuthor

    Route corrected — this one is not a resync case

    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 worked here in any case: this is a structural-parity failure, and a retry reproduces those byte-for-byte (verified on the fr numpy.md pair, where the original run's issue and the resync retry's issue have identical bodies at two different engine versions).

    Status. lectures/learning_approximation.md is absent from lecture-python.zh-cn — no target file and no .translate/state entry — against a source last touched at a4fbd865. The edition reads 143/145.

    Route. translate init -f learning_approximation from a source clone pulled to origin/main. init reads the source document off the local filesystem and never consults a merge SHA, so it cannot exhibit the #276 stale-snapshot defect, and its parity check is anchored to current main.

    Blocker unchanged. action-translation#118 still gates this — and the primary evidence for that block is this issue and its sibling, whose bodies carry the failing run verbatim (directive count differs: source has 35, output has 34). Until #118's underlying generation defect is addressed, init will refuse to write rather than produce a corrupt file, which is the v0.24.0 parity guard behaving correctly.

    Tracked target-side at lecture-python.zh-cn#210.

  2. mmcky commented on Aug 20, 2026

    @mmcky
    ContributorAuthor

    Recovered and closing. learning_approximation.md is on lecture-python.zh-cn@main as of zh-cn#261 (merged 2026-08-20). That edition's coverage is now 144/145.

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

    Two things worth recording for action-translation#118, which is still open:

    The parity failure reproduces on the init path, so it is not sync-specific — all three write paths hit it.

    But it is not deterministic. Across six attempts the run refused four times and succeeded twice, always with the signature in this issue's body: the output drops the leading ```{raw} header block, so directive #1 shifts {raw} → {index} and the count lands one short (36 vs 35). The guard is fail-closed and wrote nothing on any refusal. That is a narrower and more diagnosable target than "wholesale mismatch", and it means a retry genuinely can succeed on this path — unlike the sync path, where the recorded retry reproduced its failure byte-for-byte.

    The merged output was verified after the fact rather than trusted: 36/36 directives, 17/17 code cells, 15/15 headings, 0 python syntax errors across all cells, 4/4 figure captions localised with none byte-identical to source.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions