tech-toolchain archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Is __STDC_VERSION supposed to be defined in NetBSD?

On Jul 30,  9:18am, Hisashi T Fujinaka wrote:
} On Wed, 30 Jul 2008, Alan Barrett wrote:
} > On Tue, 29 Jul 2008, Hisashi T Fujinaka wrote:
} >> Is __STDC_VERSION supposed to be defined in NetBSD?
} >
} > I don't see why it should be.  However, __STDC_VERSION__ is defined
} > if you run the compiler in c99 mode.
} >
} > $ cpp -dM -std=c99 </dev/null | grep STDC
} > #define __STDC_HOSTED__ 1
} > #define __STDC_VERSION__ 199901L
} > #define __STDC__ 1
} Ah. Maybe that's what jnemeth meant in his comment to the pr. There was
} a cryptic ed command at the end of it.

     Yeah, when I said __STDC_VERSION, I meant to say __STDC_VERSION__.

} So is this back to a pkg pr where the Makefile needs the -std=c99?

     I personally still feel that it is a toolchain issue; however,
I've been testing a possible fix for the sendmail package.
Unconditionally setting -std=c99 will fail on systems that don't
provide a C99 compiler.

     The big issue is that in -current some headers will pull in
<stdbool.h> even without -std=c99 (this is a change in behaviour
from 4.0).  This means that bool is defined differently from how
sendmail defined it.  sendmail keys on __STDC_VERSION__ to determine
if <stdbool.h> exists.  If __STDC_VERSION__ is defined then sendmail
will pull in <stdbool.h> instead of providing its own definiton of
bool.  This is probably the best it can do without doing autoconf
style stuff.

}-- End of excerpt from Hisashi T Fujinaka

Home | Main Index | Thread Index | Old Index