This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: g++.old-deja/g++.mike/p7325.C - suspected bogus test case
- From: Mike Stump <mrs at apple dot com>
- To: Mark Mitchell <mark at codesourcery dot com>
- Cc: Zack Weinberg <zack at codesourcery dot com>, gcc at gcc dot gnu dot org
- Date: Tue, 29 Jun 2004 07:30:48 -0700
- Subject: Re: g++.old-deja/g++.mike/p7325.C - suspected bogus test case
On Saturday, June 26, 2004, at 12:37 PM, Mark Mitchell wrote:
I can certainly think of situations where the compiler could fail this
test with optimization enabled
This test cases doesn't use optimization, and test case that did, would
be a totally different test case. We aren't talking about a different
test case, we are talking about the exact one mentioned.
So, I agree -- the compiler could reasonably fail this test either
with optimization enabled or disabled, on plausible architectures.
I disagree. To prove this, fix the test case, and see if it fails.
You'll discover that in fact the test case is reliable and does work.
I'd rather revisit this path when someone wants to put in an
optimization that causes the testcase to fail, we can then see if there
are ways to ensure that it works, or find ways that it will work.
Removing the test cases because there might exist a future optimization
that might make it not work, is, well, premature in my book.
For example, if someone wanted to consume more stack to make the test
case execute faster, than I would say add -Os to the testcase and
ensure that the compiler still reuses the stack space.
Therefore, it's OK to remove this test for 3.4.1 and on the mainline.
:-( I object.