This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch: changes to patch_tester.sh
- From: "Sebastian Pop" <sebpop at gmail dot com>
- To: "Tom Tromey" <tromey at redhat dot com>
- Cc: "Gcc Patch List" <gcc-patches at gcc dot gnu dot org>
- Date: Wed, 5 Mar 2008 16:58:20 -0600
- Subject: Re: Patch: changes to patch_tester.sh
- References: <18383.4432.278950.22147@fleche.redhat.com>
On Wed, Mar 5, 2008 at 3:32 PM, Tom Tromey <tromey@redhat.com> wrote:
> In an effort not to miss regressions any more, I've started using
> patch_tester.sh on the gcc compile farm.
>
> This patch is a grab bag of a few different fixes and changes:
>
> * usage() is wrong about the watermark argument
>
> * Now you can set up a tester with a $STATE/defaults file, which has
> system-specific default settings for configure, make, and make
> check. I use this on the compile farm to pass --with-mpfr.
>
Janis has also some patch for this, but she did not posted her changes
yet. This is really needed for portability between several systems.
> * I changed selfexec to use $0 since I wanted the tester to always
> restart my hacked version. This change could go either way, I
> guess... once the other bits are in I suppose I probably would not
> need this, at least not until I made another change.
>
> * If a patch removes all the contents of a file, we now delete the
> file. I wanted to test a patch that did an svn rm and this was
> needed. I think it is more likely that this scenario means that the
> file was deleted than that the user really wanted a zero-length
> file.
>
> FWIW patch_tester.sh already doesn't work with the mercurial
> repository, AFAICT. So, I think my patch hack didn't really break
> anything here. In the future, assuming we do switch to hg, I'd
> assume we can play some fun DVC game instead of schlepping patches
> around.
>
>
> Ok?
>
Yes, the patch looks good. Thanks,
Sebastian