On 01/20/2011 04:34 AM, Jim Meyering wrote:
> This is in preparation for hooking this rule to "check"
> at least in coreutils.  Today I was burned once again by
> pushing a non-public submodule reference (coreutils).
> The fact that this check is currently run only at release time
> is good, but obviously not enough.
> 
> It probably belongs on a server-side commit hook, too.
> 
> As a possible follow-on idea, ...
> So far, I don't see a down-side to adding this line to maint.mk:
> 
>   check: public-submodule-commit

The rule is a no-op for projects that use gnulib and maintainer-makefile
but do not use git submodules (or even git at all), so I'm okay with it
on that front.

However, there are times when I specifically commit a private gnulib
commit to my private clone of a repository, since that makes it easier
to pull from my private clone on multiple machines to test whether a
potential gnulib patch has merit.  I guess it would be easy enough to
use 'make -k check' in those cases, although it becomes a bit more
effort to inspect the logs in those cases to ensure the only failure was
indeed from the known-private submodule and not from a regression
introduced by using the candidate gnulib commit.

I'm 60:40 in favor of including the 'check: public-submodule-commit'
hook at the moment.

-- 
Eric Blake   [email protected]    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to