WhoamiI00 commented on issue #44403:
URL: https://github.com/apache/superset/issues/44403#issuecomment-5749471797

   I reproduced this on current `master` (`4511c138`) and the drift is still 
there, slightly wider than when the issue was filed:
   
   | | at `c426186a2e` (issue) | at `4511c138` |
   | --- | --- | --- |
   | committed `messages.pot` | 5117 | 5123 |
   | fresh `pybabel extract` | 5436 | 5445 |
   | in source, **not** in template | 428 | **446** |
   | in template, no longer in source | 109 | **124** |
   
   All the examples quoted in the issue are still absent from the template, so 
the finding holds.
   
   **One thing worth flagging before anyone regenerates: this needs to land 
*after* #44395, not before.**
   
   #44395 fixes `babel_update.sh`'s normalization step, which currently fails 
silently — `msgcat --sort-by-msgid` is not a valid option, and with no `set -e` 
the script carries on and exits 0. The effect is that `--no-wrap` has never 
been applied, so the committed template is Babel-wrapped.
   
   That matters here because the two changes touch the same file in 
incompatible shapes. Regenerating today produces a wrapped template; #44395 
then rewrites the wrapping of every entry it touches. Doing it in that order 
means a large avoidable conflict and a second churn of the whole file.
   
   They are genuinely complementary rather than duplicates, which I checked 
rather than assumed — I measured the drift **on top of #44395's branch**:
   
   ```
   committed messages.pot : 5123
   fresh extraction       : 5445
   in source, NOT in pot  : 446
   in pot, NOT in source  : 124
   ```
   
   Identical to master. #44395 is +461/−457 msgids, i.e. a re-wrap rather than 
an extraction refresh, and none of the strings this issue names appear as 
additions there. So #44395 fixes the *shape* of the template and leaves the 
*contents* stale; this issue is the contents.
   
   Happy to do the regeneration once #44395 merges, running 
`scripts/translations/babel_update.sh` as-is (with the fixed `msgcat` line) 
rather than a hand-rolled extraction, so the `.po` catalogs, the 
do-not-translate stamps and the trailing-blank-line handling all come out as 
the script intends. Worth pinning Babel to 2.17.0 while doing it, to match the 
committed `Generated-By` and keep the diff to real string changes instead of 
formatter churn between versions.
   
   Happy to leave it to @glaterza instead if they would rather fold it into 
their i18n work — they are clearly deep in this area right now.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to