On Thu, Aug 07, 2014 at 05:10:43PM +0000, Joseph S. Myers wrote:
> On Thu, 31 Jul 2014, Marek Polacek wrote:
>
> > A few issues: currently we don't warn (in -std=c90 -pedantic mode)
> > about the usage of __func__ and _Pragma (possible others - va_copy?).
> > I'm hoping this is not that important though. For restrict/inline,
> > in GNU90 mode we just don't compile the code at all (unless we use
> > __restrict and similar), and -Wc90-c99-compat does not warn about those.
>
> And, -pedantic -Wno-c90-c99-compat - outside C99 mode - doesn't disable
> the relevant warnings (I'd think of it as logically like -pedantic
> -Wno-long-long, so it should disable them, while leaving all the -pedantic
> diagnostics for features that aren't in C99). Though I think implementing
> that could reasonably be a followup fix.
>
> E.g., in C90 mode, -Wlong-long is more specific than -Wc90-c99-compat,
> which is more specific than -pedantic. So the "long long" warnings should
> be determined by any explicit -W(no-)long-long if given, otherwise by any
> -W(no-)c90-c99-compat if given, otherwise by -pedantic, following the
> general rule of the most specific option taking precedence (and
> command-line order only being relevant when multiple variants of the same
> option are used, e.g. -Wlong-long -Wno-long-long). Outside C90 mode,
> -pedantic is irrelevant to such warnings; -W(no-)long-long should take
> precedence over -W(no-)c90-c99-compat if both are used. For warnings
> without a more specific option such as -Wlong-long, of course it's a bit
> simpler.
Yeah, makes sense. I just finished testing a patch that implements
the above; will post it momentarily.
> "a specific option".
>
> OK with that change.
Fixed, thanks for reviewing this.
Marek
