[MAILER-DAEMON@sourceware.cygnus.com: failure notice]

[Shiv Shankar Ramakrishnan]

|2.  http://www.dinkumware.com/htm_cpl/map.html#map::erase

Well it seems that dikumware has got it wrong as far as the iterator
erase() for maps go. Apparently the claim is that those online docs are
in sync with the published ISO std. My documentation of the dinkumware
libs with VC++ also shows the wrong return types for erase and indeed
the header does have the same signature. So much for the std compliance
claim by dinkumware. What is more funny is that the RogueWave
implementation that he looks down upon has got it correct :)(looking at
the borland headers).
Was the signature different in the draft C++ std that dinkumware might
be following it? 
Thanks,
Shiv