Commit c42be2c
Build strict: enable -W in ci.yml and publish.yml
Both builds carried `# TODO: Re-enable -W flag once all lectures are
translated and warnings are resolved`. The precondition is already met:
this edition has 26 of 26 lectures with a `_toc.yml` identical to
source, so there are no untranslated lectures left to produce dangling
cross-references.
The case for strict is stronger than build hygiene. A translated
edition carries the same code as its English source, so a warning here
that does not appear upstream is evidence of a translation defect, not
noise — which makes -W a translation-quality signal.
It is also the last line of defence against a class of silent engine
corruption. QuantEcon/action-translation#118 (a resync wrapped an
entire 1,465-line document in a code fence and mis-derived its title
from a Python tuple-unpacking line) and #119 (a `{raw} jupyter`
directive argument stripped to `{raw}`) both reported success at run
time, looked plausible in review, and surfaced weeks later only
because a downstream repo built with -n -W. Without the flag this
edition would merge that class of damage green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>1 parent 40df73d commit c42be2c
2 files changed
Lines changed: 2 additions & 4 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
42 | 42 | | |
43 | 43 | | |
44 | 44 | | |
45 | | - | |
46 | | - | |
| 45 | + | |
47 | 46 | | |
48 | 47 | | |
49 | 48 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
56 | 56 | | |
57 | 57 | | |
58 | 58 | | |
59 | | - | |
60 | | - | |
| 59 | + | |
61 | 60 | | |
62 | 61 | | |
63 | 62 | | |
| |||
0 commit comments