[PATCH] Support printing inlining call stack in diagnostic messages

Jason Merrill jason@redhat.com
Fri Sep 21 07:24:00 GMT 2007


Jakub Jelinek wrote:
> I tried to match what GCC already does with the "included from" stuff
> In file included from foo.h:2,
>                  from foo.c:2:
> bar.h: In function 'foo':
> bar.h:5: warning: call to __builtin___memcpy_chk will always overflow destination buffer
> messages, with instantiated from messages etc.
> The virtual call stack really belongs before the warning and should
> not be a note - if you have several warnings or errors in the same inlined
> function, it is something that should be printed just once, not after
> every such a message.

I agree with this.  The inline context is part of the context just like 
the include path and template instantiation context.  Where did Mark 
weigh in on this question?

Jason



More information about the Gcc-patches mailing list