Description
Head of main
Looking only at the translations linked in our language picker, I see these numbers right now using main
:
pt-BR before: 1597 translated messages, 196 fuzzy translations, 727 untranslated messages.
zh-CN before: 1634 translated messages, 193 fuzzy translations, 693 untranslated messages.
zh-TW before: 1505 translated messages, 137 fuzzy translations, 878 untranslated messages.
ko before: 1968 translated messages, 204 fuzzy translations, 348 untranslated messages.
es before: 2018 translated messages, 112 fuzzy translations, 390 untranslated messages.
This is after updating the translations with the newest strings from main
:
MDBOOK_OUTPUT='{"xgettext": {"pot-file": "messages.pot"}}' mdbook build -d po
for lang in pt-BR zh-CN zh-TW ko es; do
msgmerge --update po/$lang.po po/messages.pot
echo -n "$lang before: "; msgfmt -o /dev/null --stat po/$lang.po
done
Head of cr2-organization
Updating to 9dcb262 on cr2-organization
and running the same commands will show the impact of the branch in terms of messages that have been touched. I get these numbers:
pt-BR after: 1086 translated messages, 351 fuzzy translations, 1148 untranslated messages.
zh-CN after: 1160 translated messages, 320 fuzzy translations, 1105 untranslated messages.
zh-TW after: 1019 translated messages, 281 fuzzy translations, 1285 untranslated messages.
ko after: 1453 translated messages, 377 fuzzy translations, 755 untranslated messages.
es after: 1530 translated messages, 274 fuzzy translations, 781 untranslated messages.
In table form:
Language | Translated before | % | Translated after | % |
---|---|---|---|---|
pt-BR | 1597 | 63% | 1086 | 42% |
zh-CN | 1634 | 65% | 1160 | 45% |
zh-TW | 1505 | 60% | 1019 | 39% |
ko | 1968 | 78% | 1453 | 56% |
es | 2018 | 80% | 1530 | 59% |
This suggests that the new material touch about 500 messages, or 1/3 of the content.
Using back-dated sources
Now, the above shows the raw delta between main
and cr2-organization
. The actual impact on the translations will be different: the translations are published using back-dated sources, sources that match the time of the last msgmerge
.
The statistics for the translations right now (using back-dated sources) is:
Language | Translated now | % |
---|---|---|
pt-BR | 1762 | 71% |
zh-CN | 1658 | 67% |
zh-TW | 1644 | 66% |
ko | 2175 | 88% |
es | 2101 | 86% |
Risks
When we introduced the snapshot publishing system in #1243, we also created a risk that a translation will stay behind forever.
I wonder if translations will chose to not jump past the merge point with cr2-organization
?
We can mitigate this by having our internal Localization team work on updating the translations after the merge.