Is there some reason gettext is special here?
Yes. At least in Bruno's time: he was adamant that the gettext files
come as a group from an official release because he did not want partial
updates of "some" files coming through. It seemed reasonable when he
said it.
(Aside: the whole gettext-in-gnulib concept has always induced massive
confusion in my brain when I'm trying to update gettext. The gnulib
documentation about it just increases my confusion. I somewhat wish
that gettext only existed on its own, and was not part of gnulib at all,
which would completely clarify what comes from where. Anyway.)
I didn't know that it was supposed to be the tarball release for
gettext -- that's not the practice for the other upstream sources.
Not exactly. The only other "upstream" remaining which makes tarball
releases is automake, and gnulib only imports standalone shell scripts
from there, hence doesn't have to be from a release. (Oh, texinfo.tex
too; same story there.) In contrast, the gettext files are a coherent
group, not individual standalone files.
no real harm in this particular case.
The harm is that my checking script notices the discrepancy every day,
and I don't want to see it.
Either the discrepancy should exist, or it should not. Meaning that
either the files should come from development gettext, or from released
gettext. Which way is better is up to Daiki or you or someone other
than me, but I'd like the situation to be consistent ... -k
