This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Fixing libstdc++/3628 (was: Re: GCC 3.1 Release)
--On Thursday, April 11, 2002 12:48:45 AM +0200 Paolo Carlini
<pcarlini@unitus.it> wrote:
> Phil Edwards wrote:
>
>> How about 20_util/rel_ops.cc ?
>>
> [...]
>
>> By the way, the changelog entry only gets sent to GNATS if the line reads
>>
>> PR libstdc++/3628
>>
>> i.e., with the leading "PR " text. I just noticed that none of
>> Benjamin's recent fixes have that text either. Oh well. :-) No easy
>> way to make it work retroactively.
Gaby, I understand your objection here, I think. You're saying that since
we're creating these new operators in namespace std, saying:
"using std::operator=="
will pull in more things that it should. Can a conforming program
detect that difference?
I agree that putting __normal_iterator and friends into a separate
namespace would be a cleaner solution. It sounds pretty invasive,
though. Do you think it's practical to do that quickly?
If not, I'd suggest we go with Joe/Paolo's patch, and then file a PR
about the "using std::operator==" issue. I think that would probably
be a step forward. Remember that sometimes we can't do everything
we'd like to, but we still want to make progress...
What are your thoughts on that?
Thanks,
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com