PR 10692

Paolo Carlini pcarlini@unitus.it
Mon Jul 14 22:13:00 GMT 2003


Mark Mitchell wrote:

>Do any of the V3 people know anything more about 10692?
>
>This is a regression on Solaris, which is a primary platform for GCC.
>  
>
10703, you mean?

In that case, it affects Solaris but only 2.5.1 and 2.6, not 2.7-9... 
Something is definitely wrong with the OS provided 'fstat' or 'poll'. 
Despite the succesful autoconf tests, one of those misbehave at run time 
(the code affected is that in __basic_file<>::showmanyc 
(libstdc++-v3/config/io/basic_file_stdio.cc)). The finer grained tests 
present in the 3.4 branch reveal that it's only 
27_io/basic_istream/readsome/char/6746-2.cc to fail. I don't consider 
this a very serious failure.

I'm afraid that without some consistent help from Solaris people there 
are meager chances to have the problem worked around (I don't see how it 
can possibly be actually -fixed-) any time soon...

About 11346, I have no idea at the moment...

Paolo.



More information about the Libstdc++ mailing list