Re: [Quilt-dev] test suite improvements

[Andreas Gruenbacher]

On Sunday 15 January 2006 07:54, John Vandenberg wrote:
> Andreas,
>
> The series of patches you were looking through were very old, created
> to get the MSYS port operational.  I have been redoing them all again,
> from scratch.
>
> The patches that I proposed last week for 0.43 are here:
>
> http://zeroj.hda0.net/quilt-0.43-patches/
>
> The first part of that series is mostly redundant since your commits,
> and the rest is stale.  I have refreshed all of the useful components,
> except for the 'reference.diff' patch.  It takes a different approach
> to the building of quilt.1 and README, improving the build times
> considerably.  Is it worth me refreshing that patch for 0.43?

I don't think we need a --reference option, but the result is almost the same 
as the reference target right now, so if you think that building is still 
slow, you could do something like:

        doc/reference: bin/quilt $(QUILT:%=quilt/%) scripts/patchfns
                $(MAKE) -s reference > $@

On Linux this is already so fast it doesn't matter.

> Here is an updated series against trunk:
>
> http://zeroj.hda0.net/quilt-0.43-patches-v2/

I'll have another look, thanks.

> I forgot to mention, that patch also changes the output of failed cases
> from
>
> <expected>  <sep>  <actual>
>
> to
>
> <actual>  <comparitor>  <expected>

Okay.

> There is minor issue with your changes: symlinks are not created in
> compat/ , causing make check to fail when configure
> --with-<name>=<prog> is used.  the attached patch resolves that; is it
> ok for me to commit it?  (btw, thanks Jean for the bash tip)

Thanks, I have just submitted it after fixing up several other things.

> > http://zeroj.hda0.net/quilt/patches/test-msys.diff:
> > Please use $(filter-out some.test,$TESTS) instead of patsubst. I don't
> > think it's great to hardcode "MINGW32" in anything but configure.in, but
> > in this case it might be better than the alternatives.
>
> I'll write two new autoconf macros to check whether the platform
> supports POSIX file permissions, and if patch -e works.

Okay.

Thanks,
Andreas