Julian Foad wrote:
> Branko Čibej wrote:
>
>> I thought the gettext tools could verify that the format strings in
>> the translated messages are correct? If they can, we should do this
>> checking in a post-commit hook.
>
>
> That's an interesting idea, doing build checking in a post-commit
> hook*. You mention it casually as if it's the normal thing to do, but
> I'm not aware that we currently do any such thing.
It's a normal thing for me. :)
> Is our existing practice of relying on committers to do check their
> commit (e.g. "make" and "make check" for core C code), and the
> automatic periodic running of regression tests, not appropriate?
It is, if you substitute "developers" for "committers". We're talking
about translations here, and we can't require translators to build and
test all of Subversion, and even if we did, the format checking (which
only GCC knows about, other compilers don't, sigh...) happens at compile
time, not on the translated strings. :)
I don't think that running "msgfmt --check-format" on the .po file
before checkin is too hard, but on the other hand, it's not too much to
do in a post-commit hook. Note that we already do this in the reguar build.
(Hah, but not in the Windows build, which I'm fixing now.)
> If not, can we make use of this post-commit idea for all sorts of
> tests, not just checking the format of translations?
Only for things we can reasonably expect svn.collab.net to handle.
Regression testing isn't one of them. Current regression testers are
already free to hook into the commit e-mails if they want triggers.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Sep 3 14:50:14 2005