[Proposed binutils PATCH] Re: Diagnosing an intricate C++ problem

[Loren James Rittle]

In article < 20000902170839.A1895@lucon.org >,
"H . J . Lu" <hjl@lucon.org> writes:

> I guess I misunderstood. I don't like this patch. It creates way too
> many false positives. And I don't see how linker can tell them apart.

As the author of the original patch, I completely agree.  The patch
posted some time ago was for Gerald and others to comment upon.  Given
the false positive rate seen when all statically-linked code is not
compiled with the exact same flags, it could not be made the default
in it current form.

Thank you for your comments.  Would you mind the feature going into ld
as long as it were an option that must be explicitly enabled by C++
programmers that desire it?

Regards,
Loren