Solaris package, was Re: Mercurial 1.3 released!

[rupert.thurner]

On 29 Jul., 22:32, Martin Geisler <m... at lazybytes.net> wrote:
> Mads Kiilerich <m... at kiilerich.com> writes:
> >http://www.selenic.com/hg/index.cgi/rev/6e5e548452de(test-gpg:
> > suppress secure memory warning) should probably have been applied to
> > stable as well. Anyone?
>
> I've transplanted it to crew-stable (b3ee4c2e1ff5).
>
> > Regarding test-repair-strip: One of the patches from
> >http://www.selenic.com/pipermail/mercurial-devel/2009-July/thread.htm...
> > should be applied to stable.
>
> In crew-stable as 4a1b24dbf753, thanks again.

thanks for this, gpg, transplant, and repair-strip errors are gone!

i tried now the crew-stable-53fdf18fd63b source checkout, and also put
the gnu tools in the path to have a gnu diff. cvs version is 1.11.23.
i run it on multiple platforms and operating system versions, parallel
and non-parallel.

 0. solaris8, sparc, mercurial-1.3 (old test): http://pastebin.com/m32e3576f
 1. solaris8, sparc, mercurial-1.3.1: http://mercurial.pastebin.com/m47cfc5f9
 2. solaris8, sparc, crew, parallel: http://pastebin.com/m5f53866c
 3. solaris8, sparc, crew: http://mercurial.pastebin.com/m9b2365d
 4. solaris10, i386, crew, parallel: http://mercurial.pastebin.com/mdcf6ef7
 5. solaris10, sparc, crew: http://mercurial.pastebin.com/m277a964e
 6. solaris10, sparc, crew, parallel: http://pastebin.com/m4e3a4ede
 7. solaris10, sparc, crew, parallel: http://mercurial.pastebin.com/m14098f50

the parallel option reduces the test runtime from 63 minutes to 7
minutes, on a sun t5220. strange is that the runs with -j can differ
quite a lot, while the runs without -j seem to be consistent, no
matter if we do it from source or with the http//opencsw.org gar build
mechanism.

rupert.