[Patchlet/RFC] To basic_string::_Rep::_M_destroy

[Paolo Carlini]

Hi,

this is something I have been able to notice only thanks to the
ongoing restructuring work: basically, when _M_destroy is called,
either it's a "real" string, dynamically allocated, that is (case
of the _S_construct(s)) or the caller is _M_dispose, which already
checks at the outset for _S_empty_rep.

In the branch, this simplification is not giving any problem, and
since seems so trivial, I'd like to apply it to mainline too... if
nobody spots something wrong with it, indeed.

I'm finishing regtesting on x86/x86_64-linux.

Paolo.

///////////
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: CL_destroy
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050222/d9f17f24/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_destroy
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050222/d9f17f24/attachment-0001.ksh>