Re: 3.0 sparc-sun-solaris2.8 Regression: c_math.cc Now Fails

[Jeffrey Oldham <oldham at codesourcery dot com>]

On Sun, May 27, 2001 at 11:04:49AM +0200, Gabriel Dos Reis wrote:
> Carlo Wood <carlo@alinoe.com> writes:
> 
> | On Sat, May 26, 2001 at 03:01:17PM -0700, Jeffrey Oldham wrote:
> | > 
> | > gcc 3.0 for sparc-sun-solaris2.8 now fails on
> | > libstdc++-v3/testsuite/26_numerics/c_math.cc:
> | > 
> | > 		    === libstdc++-v3 tests ===
> | >     Running target unix
> | >     FAIL: 26_numerics/c_math.cc (test for excess errors)
> | > 
> | > It did not fail yesterday (25May).
> | > 
> | > Would someone with access to a sparc-sun-solaris2.8 computer please
> | > find the offending patch that caused this?
> 
> That may be caused by a patch I checked in yesterday which revealed a
> latent bug.  
> 
> | > Under Mark Mitchell's directive
> | > (http://gcc.gnu.org/ml/gcc/2001-05/msg01044.html) any patch that
> | > causes a regression will be summarily reverted.  Once you find the
> | > patch, you are empowered to revert it.  I can do the dirty work if
> | > desired.
> | 
> | This could be the problem with sqrt I saw today on the list?
> | In that case it was also already fixed today.
> | The subject of the mail with the patch was something like:
> | V3 PATCH: #undef abs ...
> | I suppose the 'abs' was a typo, it should have been #undef sqrt ;)
> 
> Definitely.  

Great!  I assume this resolved.  (The nightly sparc-sun-solaris2.8
build failed for other reasons today (27May).)

Thanks,
Jeffrey D. Oldham
oldham@codesourcery.com