<URL: http://bugs.freeciv.org/Ticket/Display.html?id=39976 >
There have been public and private reports about directly adding new msgid,
when there is an old msgid with the same text. Apparently (although not
in my experience), there's a make error.
Ideally, the code would match the old msgid with the new one, and compare
(1) When the msgstr is identical, and neither of the entries is fuzzy, it
should just ignore the old comment.
(2) When the msgstr is identical, and one of the entries *is* fuzzy, it
should just ignore the old comment, but ensure that fuzzy is marked.
(This is the usual case, indicating a potential new usage.)
(3) When the msgstr is different (as is usually the intent), it should
just add the old (comment) version immediately below the new version for
easy comparison by the translator, and mark the whole thing fuzzy!
In the most recent example, there was an old "_Research" from gui-mui.
Since I do most of my work on S2_2 and backport to S2_1, there's no
gui-mui, so the old string didn't show up in the C source.
Since gui-mui is still present in S2_1, but at some time stopped being
indexed for po purposes, the old strings were moved to comments. That's
great. But the scripts aren't up to the task.
Again I ask, are these the most recent po scripts?
Freeciv-dev mailing list