debug/vector anomalous behaviour

[PAOLO.CARLINI@ORACLE.COM]

> Right, they can only know what capacity is guaranteed by the
> standard, which either grows minimally (when inserting) or is set by
> reserve().

Great. I'm attaching the patch as actually tested... Assuming there are no objections, I will apply it to mainline and for now revert the first try of yesterday from 4_3-branch: better play safe in the release branch, the issue isn't a regression and, as strange as it seems, nobody reported it to date.

Paolo.

//////////
-------------- next part --------------
A non-text attachment was scrubbed...
Name: CL_dmrr_2
Type: application/octet-stream
Size: 452 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20080511/965558c3/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch_dmrr_2
Type: application/octet-stream
Size: 4536 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20080511/965558c3/attachment-0001.obj>