PR 36778: Should we have -Wfatal-warnings

[Paul Brook]

> Hi,
> 
> I was having a look to this long standing, and unconfirmed, PR:
> 
>      http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36778
> 
> and the rationale makes sense to me. What do you think, shall we have
> -Wfatal-warnings too, together with -Wfatal-errors?
> 
> AFAICS, the patch would be rather trivial, little more than the
> attached, modulo a few tests and a few lines of documentation...

This isn't actually necessary, they just need to make their test harness a bit 
smarter.  

If gcc promotes a warning to an error then it prints

  "cc1: all warnings being treated as errors"

If you see this message then you hit a warning (promoted to an error).
If you don't see this message then you hit an actual error.

A quick test shows this is true for all gcc since at least 4.4.

Paul