Re: lib/regex_internal.h on 1.4.11

On Wed, Oct 08, 2008 at 02:06:28AM +0200, Bruno Haible wrote: > Eric Blake wrote: > > > Indeed, we're in a maze of twisty little HP cc compiler bugs: > > > > > >

Re: lib/regex_internal.h on 1.4.11

Eric Blake wrote: > > Indeed, we're in a maze of twisty little HP cc compiler bugs: > > > > > > minus_zerol is miscompiled on at least Irix 6.2. So until w

Re: lib/regex_internal.h on 1.4.11

On Mon, 6 Oct 2008, Eric Blake wrote: minus_zerol is miscompiled on at least Irix 6.2. So until we hear feedback from Jonathan on what DOES allow us to compile -0.0L on HP-UX/hppa, I'd rather go for consistency in the testsuite by always using - -zerol as the common idiom. I don't know if I'm

Re: lib/regex_internal.h on 1.4.11

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 9/13/2008 2:52 AM: >> Meanwhile, I have a question about tests/test-signbit.c - your >> patch changed most occurrences of -0.0L to -zerol, to work around one HP >> bug, but there is another comment that states: >> >> /* We

Re: lib/regex_internal.h on 1.4.11

Eric Blake wrote: > It looks like you failed to check .m4 tests. I'm applying the patch > below. Yes, I missed to change the .m4 tests. Thank you for the fix. > Meanwhile, I have a question about tests/test-signbit.c - your > patch changed most occurrences of -0.0L to -zerol, to work around one

Re: lib/regex_internal.h on 1.4.11

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 8/1/2008 3:30 AM: > Jonathan C. Patschke wrote: >> That fixes the problem in test-frexp and test-strtod, but not test-frexpl >> (which still fails the (mantissa == x) assertion). > > Thanks. I'm applying the patch below, a

Re: lib/regex_internal.h on 1.4.11

Jonathan C. Patschke wrote: > That fixes the problem in test-frexp and test-strtod, but not test-frexpl > (which still fails the (mantissa == x) assertion). Thanks. I'm applying the patch below, assuming that if the compiler confuses -0.0 with 0.0, it also confuses -0.0L with 0.0, and -0.0f with 0

Re: lib/regex_internal.h on 1.4.11

On Thu, 31 Jul 2008, Bruno Haible wrote: test-frexp.c:103: assertion failed sh[9]: 15807 Abort(coredump) FAIL: test-frexp test-frexpl.c:114: assertion failed sh[9]: 15824 Abort(coredump) FAIL: test-frexpl test-signbit.c:55: assertion failed sh[9]: 16478 Abort(coredump) FAIL: test-signbit test-st

Re: lib/regex_internal.h on 1.4.11

On Jul 31, 2008, at 3:42 AM, Bruno Haible wrote: Jonathan Patschke wrote: | The literal constant on line 164 of lib/regex_internal.h needs to be | explicitly declared as an unsigned long. HP-UX's preprocessor (at least | on 10.20) will fail a comparison between 0x and 0xU

Re: lib/regex_internal.h on 1.4.11

Jonathan Patschke wrote: > > If you change that line to > > #if BITSET_WORD_MAX == 0xUL > > and recompile, do the compilation and "make check" then succeed? > > [... test results with nothing wrong regarding regular expressions...] Thanks for testing; I'm applying this fix (and also fixin

Re: lib/regex_internal.h on 1.4.11

Jonathan Patschke wrote: > | The literal constant on line 164 of lib/regex_internal.h needs to be > | explicitly declared as an unsigned long. HP-UX's preprocessor (at least > | on 10.20) will fail a comparison between 0x and 0xUL. > | The constant in limits.h is explicitly UL, so

Re: lib/regex_internal.h on 1.4.11

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Jonathan Patschke on 7/30/2008 5:55 PM: | | The literal constant on line 164 of lib/regex_internal.h needs to be | explicitly declared as an unsigned long. HP-UX's preprocessor (at least | on 10.20) will fail a comparison between 0xf