Re: Bug#410537: build failure on alpha
On Sun, Feb 25, 2007 at 03:00:04PM -0600, Steve M. Robbins wrote:
> Thanks hugely for looking into this.
> On Sun, Feb 25, 2007 at 10:36:21AM -0800, Steve Langasek wrote:
> > On Sun, Feb 25, 2007 at 01:14:44AM -0800, Steve Langasek wrote:
> > > Looking at the code, I gather that it's expected for this test to throw an
> > > assertion, and the problem is that it blows up on the other end. Other
> > > tests that use CPPUNIT_ASSERT_ASSERTION_FAIL() seem to work ok, though, so
> > > it doesn't seem to be a general problem with exception handling on alpha.
> Have you actually seen a test using CPPUNIT_ASSERT_ASSERTION_FAIL()
> work ok?
Yes, I looked for other occurrences of this in the code and tested them
directly, invoking the individual tests by name from cppunittestmain.
> Just in case you were mistaken about seeing CPPUNIT_ASSERT_ASSERTION_FAIL
> work correctly, I'm attaching a small test case to exercise it. If it
> fails to work with -O2, but does work with -O0, perhaps the assembly
> might clue us in on what is being optimized away.
That test case works fine when built with -O2, regardless of the options
used for building cppunit. I.e., even when building cppunit itself with -O2
(and getting the testsuite failure), this test case still passes.
Time allowing, I'll look into getting a minimal test case working back from
the one that's known to fail.
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.