Re: c++ implies -fno-builtin -- why??

[Paul Koning <pkoning at equallogic dot com>]

Excerpt of message (sent 12 July 2002) by Roger Sayle:
> 
> > In digging through the docs I ran across the statement that C++
> > ignores -fbuiltin and always runs with -fno-builtin.
> 
> This is historical.  GCC builtin functions were temporarily disabled
> for the GCC 3.0.x releases of g++ to solve C++ conformance issues.
> For example, with the 2.95 compiler g++ didn't issue a diagnostic
> if a user called std::strlen without first prototyping it.  These
> issues were fixed in time for GCC 3.1.x, which re-enabled builtins.
> 
> There is of course the possibility that not all of the documentation
> was updated to reflect the fact that c++ now recognizes builtins.
> Where did you come across the statement?

It's in the info file for gcc, where the -fno-builtin switch is
described.  The statement I mentioned is still there as of the CVS
mainline of a month or so ago (the last time I did a build where I
installed info files) so if that restriction went away in 3.1.0, the
documentation update was missed.

      paul