This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFA/testsuite] Remove all remaining gdb_suppress_entire_file
- From: Daniel Jacobowitz <drow at false dot org>
- To: Mark Kettenis <mark dot kettenis at xs4all dot nl>
- Cc: brobecker at adacore dot com, gdb-patches at sources dot redhat dot com
- Date: Tue, 8 Aug 2006 16:24:20 -0400
- Subject: Re: [RFA/testsuite] Remove all remaining gdb_suppress_entire_file
- References: <20060721002619.GE1499@adacore.com> <200607212205.k6LM5sNC003058@elgar.sibelius.xs4all.nl> <20060721232842.GA30038@nevyn.them.org> <20060808190052.GG24779@nevyn.them.org> <200608082010.k78KAB1h005794@elgar.sibelius.xs4all.nl>
On Tue, Aug 08, 2006 at 10:10:11PM +0200, Mark Kettenis wrote:
> Hmm, I guess what I really want is an UNTESTED if the failure was to
> be expected (for example the ADA tests if there is no ADA compiler on
> the system) and an error if something went wrong that shouldn't have
> gone wrong (an ICE from GCC on one of the testsuite code snippets).
> However, there probably isn't always a clear distinction between the
> two. For example, do we expect C++ snippets to compile on all
> systems? I suppose Joel's patch is progress; we can always tweak
> things later if we feel the UNTESTEDs are inappropriate.
Sounds good to me. We do expect most of the tests to compile on most
systems, except for those with "strange" dependencies - meaning
everything but C and C++ and anything with threads. But I don't see
a good way to capture this information that isn't more trouble than
it's really worth.
--
Daniel Jacobowitz
CodeSourcery