[Patch] libstdc++/22131

[Paolo Carlini]

Hi,

the below fixes this PR submitted by Martin. The issue in itself is
rather straightforward but adjusting consistently the library required a
few changes, because we were actively implementing something different
from the standard. A few observations:
1- As a consistency problem in the standard, the actual wording for
money_get is not as clear as that for num_get. I'm implementing a
consistent behavior but maybe a DR should be prepared.
2- In the same spirit, I'm punting on the well known consistency issue
wrt assigning vs "OR-ing" failbit at the end of the parsing in the
various facets, already discussed a bit with Martin and Nathan.
3- Also, I'm not trying to deal one-shot with DR 23 too, which finally
is making progress. I'd rather prefer waiting 'til the next meeting and
Plauger wording.

Testing x86-linux, will wait a while in case of comments. Will go in
mainline only.

Paolo.

////////////////
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: CL_22131
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050628/8af025b4/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_22131_complete
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050628/8af025b4/attachment-0001.ksh>