This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: V3 Testsuite PATCH: Disable tests on newlib
- From: Mark Mitchell <mark at codesourcery dot com>
- To: Benjamin Kosnik <bkoz at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org, libstdc++ at gcc dot gnu dot org, paul at codesourcery dot com, phil at codesourcery dot com
- Date: 07 Jan 2004 18:12:40 -0800
- Subject: Re: V3 Testsuite PATCH: Disable tests on newlib
- Organization: CodeSourcery, LLC
- References: <200312241552.hBOFqneS002164@sirius.codesourcery.com> <20040106215558.6b62efe5.bkoz@redhat.com>
On Tue, 2004-01-06 at 19:55, Benjamin Kosnik wrote:
> >Benjamin, I think the V3 testsuite needs a better way of dealing with
> >this situation. If using mkfifo is essential, then something should
> >be done on the DejaGNU side of the testsuite to make it possible to
> >XFAIL these tests on targets that do not have mkfifo. Or, if I am
> >simply missing the right incantation, I should discover it. :-)
>
> I agree, something should be done.
>
> I'd rather model this on the way that named locale support was done than
> as with your patch, however.
>
> Ie, make a function, use it instead of mkfifo in the test, and if a
> particular target can't hack it, just exit silently. I'd rather exit and
> mark the test as unsupported, but I cannot figure out how to do that,
> however.
It's different from locales in that it's a compile-time problem, not a
run-time problem. We'd have to do:
try_mkfifo (...) {
#ifdef _NEWLIB
exit(0);
#else
return mkfifo (...);
}
Is that what you want?
Also, is this really a sound testing methodology? You're suggesting
that tests *pass* even when they are (essentially) not being run. That
seems wrong; the ideal thing would be to mark them UNTESTED. XFAIL, as
I have suggested, seems like less of a lie than PASS.
--
Mark Mitchell <mark@codesourcery.com>
CodeSourcery, LLC