Re: g++ free function conflict

[BlueDoze]

Thanks for the answer

Actually my question is regarding the different behaviour between
cygnus compiler and g++ 3.2.3. and I want to check first if there's a
way to make the g++ 3.2.3 behave in this case as cygnus g++ compiler.



Paul Pluzhnikov <ppluzhnikov-nsp@charter.net> wrote in message 
news:<m3fz743jxh.fsf@salmon.parasoft.com>...
> bluedoze@yahoo.com (BlueDoze) writes:
> 
> [Please do not top-post].
> 
> > I mean I have to include this file as is, and I don't have any other choice.
> 
> Ok, one of 2 things is happening:
> - either the library you are using really *defines* 'free()' as
>   you stated in your original message; or
> - the library doesn't really define free(), but has a header file
>   with incorrect 'free()' prototype.
> 
> If the former, you need to seek advice from a guru, as the problem
> and solutions could be very non-trivial.
> 
> If the latter (which is much more likely the case), you should
> contact library provider to notify him of the bug in his header,
> and fix the header to either remove the bogus prototype alltogether,
> or to make it agree with the one in /usr/include/malloc.h
> 
> To find out which, run 'nm libfoo.* | grep " free$"' (where libfoo
> is the library you depend on). If you see this:
> 
>   00074900 W free        or
>   00074900 T free
> 
> then you have the 'library really does define free' case.
> If, OTOH, you see this:
> 
>            U free         or
>            w free
> 
> then you have the 'bogus prototype' case.
> 
> Cheers,