This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [RFC] PR libstdc++/9548 vs DR 231 (Ready)
- From: Benjamin Kosnik <bkoz at redhat dot com>
- To: Paolo Carlini <pcarlini at unitus dot it>
- Cc: libstdc++ at gcc dot gnu dot org, ncm at cantrip dot org
- Date: Tue, 4 Feb 2003 12:56:05 -0600
- Subject: Re: [RFC] PR libstdc++/9548 vs DR 231 (Ready)
- References: <3E3F855D.9080006@unitus.it>
>It would be easy to fix the specific bug, but I noted DR 231, in the
>Ready state, which is exactly about 22.2.2.2.2 p11. In my understanding
>we agreed that, in general, Ready DRs should be implemented ASAP and
>implementing DR 231, in fact, seems quite easy too.
>
>However, in these specific circumstances, I'm a bit worried since the
>output of float numbers in many different formats is going to change: a
>big impact on the behaviour of _many_ programs!
I think the current behavior is wrong, sadly. I believe the arguments
advanced in this DR are well-formed. We should match "C" behavior.
Since precision is 6 by default, this behavior will only be triggered by
those who explicitly set precision to zero. I think that's reasonable.
>Do we all agree that implementing DR 231 is the best thing to do?
I really think this should be done.
-benjamin