> BTW, since replacing memory functions within the library needs to be done very
> carefully at a very early stage to not risk having mixed calls, what would you
> say about we adding the function curl_global_init_mem() that works exactly
> like curl_global_init() but also allows the memory function callbacks to be
> set? I think it would properly prevent users from trying to set them at later
> time and thus cause havok.

That sounds like a good idea. Let me do this stuff, test it and I will
post the patch ASAP.

-Regards
Seshubabu Pasam
Received on 2004-05-10