This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: gas testsuite fixes for arm-vxworks
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- To: Zack Weinberg <zack at codesourcery dot com>
- Cc: binutils <binutils at sourceware dot org>, Nick Clifton <nickc at redhat dot com>
- Date: Fri, 1 Jul 2005 06:22:29 -0400 (EDT)
- Subject: Re: gas testsuite fixes for arm-vxworks
- References: <87k6kbo66w.fsf@codesourcery.com>
On Thu, 30 Jun 2005, Zack Weinberg wrote:
> Actually rather more than that; I got fed up with the way one has to
> edit arm.exp every time one adds a test case, so I augmented
> run_dump_test with a bunch of new features.
> It is now possible to use
> the in-.d-file annotations to specify which subtargets a test should
> run on,
Instead of taking the oportunity to sync functionality from ld
run_dump_test, where both "#target:" and "#notarget:" exists?
> and to expect diagnostic output longer than can be squoze into
> an "#error:" line.
Good one.
> Put it all together and arm.exp can be as simple
> as
>
> if {[istarget *arm*-*-*] || [istarget *xscale*-*-*]} {
> run_dump_tests [lsort [glob -nocomplain $srcdir/$subdir/*.d]]
> }
Like MMIX and to some extent CRIS does? :-)
One word about using "#name: " - don't. It just makes it hard
to find the actual test-case when it fails. Without the
"#name:" you know that it's in the corresponding .d file.
brgds, H-P