there is a bug report that the resulting flac binary, when compiled with gcc3, is bad: http://sourceforge.net/tracker/index.php?func=detail&aid=488702&group_id=13478&atid=113478 I don't have a system to try this out on right away; can anyone verify or elaborate on that? Josh __________________________________________________ Do You Yahoo!? Buy the perfect holiday gifts at Yahoo! Shopping. http://shopping.yahoo.com
On Tue, Dec 04, 2001 at 04:08:28PM -0800, Josh Coalson wrote:> there is a bug report that the resulting flac binary, when compiled with > gcc3, is bad: > > http://sourceforge.net/tracker/index.php?func=detail&aid=488702&group_id=13478&atid=113478 > > I don't have a system to try this out on right away; can anyone verify or > elaborate on that?I have gcc 3.0 installed here, I'll give it a try as soon as 1.0.2 is finished downloading. -- - mdz
On Tue, Dec 04, 2001 at 04:08:28PM -0800, Josh Coalson wrote:> there is a bug report that the resulting flac binary, when compiled with > gcc3, is bad: > > http://sourceforge.net/tracker/index.php?func=detail&aid=488702&group_id=13478&atid=113478 > > I don't have a system to try this out on right away; can anyone verify or > elaborate on that?The bug submitter is using 3.0.1; I have 3.0.2 here (Debian's 3.0.2-4). I can encode and decode a test WAV file with no problems. "make check" runs successfully for a long time before blowing up, but it breaks with both the gcc2- and gcc3-compiled versions. The error looks like this: gcc 2.95.4: noise: encode...decode...compare...cmp: EOF on noise.cmp ERROR during compare of noise FAIL: ./test_streams.sh ==================1 of 2 tests failed ================== gcc 3.0.2: noise: encode...decode...compare...noise.raw noise.cmp differ: char 660729, line 2573 ERROR during compare of noise FAIL: ./test_streams.sh ==================1 of 2 tests failed ================== I assume these are two slightly different failure conditions due to the randomness of the noise, correct? In fact, now that I try it, I get a similar error with 1.0.1, gcc 2.95.4: sine16-16: encode...decode...compare...cmp: EOF on sine16-16.cmp ERROR during compare of sine16-16 FAIL: ./test_streams.sh ==================1 of 2 tests failed ================== Perhaps this is at least partially a problem with the self-tests themselves? I don't get segfaults with the asm optimizations anymore (yay). I haven't tried 3DNOW yet. -- - mdz
I compiled with Mandrake's 3.0.1 and I've been running the test script for ages. It's been doing "noise" tests for what seems like forever, but there have been no errors [yet]. -- Asheesh. On Tue, 4 Dec 2001, Josh Coalson wrote:> there is a bug report that the resulting flac binary, > when compiled with gcc3, is bad: > > http://sourceforge.net/tracker/index.php?func=detail&aid=488702&group_id=13478&atid=113478 > > I don't have a system to try this out on right away; > can anyone verify or elaborate on that? > > Josh > > > __________________________________________________ > Do You Yahoo!? > Buy the perfect holiday gifts at Yahoo! Shopping. > http://shopping.yahoo.com > > _______________________________________________ > Flac-dev mailing list > Flac-dev@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/flac-dev > >
Yes, the tests are pretty exhaustive. Actually if you have the test_files/ directory checked out of CVS it will run even more tests. I reran all the tests again (asm optimizations on) on redhat 7.1 with gcc 2.96 and they all passed. They are also all passing on Windows with MSVC and OS X with gcc 2.95.2. Matt, there is a slight chance that your noise tests failed because your run produced the exactly perfect data to trigger something and mine didn't. But if you get it every time that is less likely. If you get it every time in nearby test points it may be certain arguments, which you can find from streams.log. Also, if you get any errors on test_bins.sh they should be totally repeatable. I will try running them again just for kicks. Josh --- Asheesh Laroia <paulproteus@technologist.com> wrote:> Finally finished running the tests on my Athlon 1Ghz on a Mandrake > 8.1 box > with GCC 3.0.1. "All 2 tests passed". I can't account for Matt > Zimmerman's GCC 3.0.2 behavior.... It took ages to finish the tests; > it > seemed it was looping forever. > > -- Asheesh. > > On Tue, 4 Dec 2001, Josh Coalson wrote: > > > there is a bug report that the resulting flac binary, > > when compiled with gcc3, is bad: > > > > >http://sourceforge.net/tracker/index.php?func=detail&aid=488702&group_id=13478&atid=113478> > > > I don't have a system to try this out on right away; > > can anyone verify or elaborate on that? > > > > Josh__________________________________________________ Do You Yahoo!? Send your FREE holiday greetings online! http://greetings.yahoo.com
Finally finished running the tests on my Athlon 1Ghz on a Mandrake 8.1 box with GCC 3.0.1. "All 2 tests passed". I can't account for Matt Zimmerman's GCC 3.0.2 behavior.... It took ages to finish the tests; it seemed it was looping forever. -- Asheesh. On Tue, 4 Dec 2001, Josh Coalson wrote:> there is a bug report that the resulting flac binary, > when compiled with gcc3, is bad: > > http://sourceforge.net/tracker/index.php?func=detail&aid=488702&group_id=13478&atid=113478 > > I don't have a system to try this out on right away; > can anyone verify or elaborate on that? > > Josh > > > __________________________________________________ > Do You Yahoo!? > Buy the perfect holiday gifts at Yahoo! Shopping. > http://shopping.yahoo.com > > _______________________________________________ > Flac-dev mailing list > Flac-dev@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/flac-dev > >