[Patch] libstdc++/21238

[Paolo Carlini]

Hi,

I'm so sorry for the false alarm: actually, the problem seems trivial,
just matter of not using a typedef provided by numpunct/moneypunct, not
necessarily provided by a specialization. Otherwise, the num_get/put and
money_get/put facets seem to me perfectly able to deal with a numpunct
(moneypunct) providing only the minimal facilities required by the standard.

While adding the instantiation tests, I noticed that in a couple of
places we were using operator!= instead of operator== on the char_types,
unnecessarily forcing the user to provide also the former (our pod_char
in pod_char_traits.h immediately noticed ;)

Tested x86-linux, will wait a bit, just in case I'm being overly
optimistic ;)

Paolo.

//////////////////
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: CL_21238
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050429/4376bde2/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_21238
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050429/4376bde2/attachment-0001.ksh>