Re: cvs commit to glibc-package/debian by jbailey
On Thu, May 01, 2003 at 01:07:11AM +0900, GOTO Masanori wrote:
> > - debian/rules: Introduce perfect_make_check_archs concept -
> > arch's listed in this variable must pass make check cleanly,
> > or the build will fail. Start off with i386, powerpc, sparc, alpha
> > and s390.
> BTW, why is it currently only for i386/ppc/sparc/alpha/s390?
The idea here is that these arch's all pass make check without any
problems. So if a failure shows up all of a sudden, it's a regression
and the resulting deb should not go into the archive. The other arch's
all have various problems that cause make check to fail. It would be
nice if glibc had a system like gcc's where each arch could mark certain
tests as ones that they expect to fail so that we could more precisely
track a regression, but drepper has rejected that idea more that one.
He has a point: the testsuite ideally should pass on every arch. For
instance, hurd-i386 doesn't pass because our signal handling sucks. But
it would suck more if we couldn't ever get another glibc into the
archive ;)
ia64 has a segfault in their testsuite that I think is fixed in 2.3.2-1.
If that's the case I will add it to the list.
I didn't investigate the failures on the other arch's. I plan to take
the time to do that, though.
Tks,
Jeff Bailey
--
You said homosexuals form a small percentage of the population. So
do Jews. Is that a reason to deny someone equality?
- Richard Marceau
Reply to: