On Thursday, November 18, 2010 22:38:18 Bruno Haible wrote:
> Mike Frysinger wrote:
> > > > The current program_invocation_name symbol detection fails if the
> > > > argp.h header is missing.
> > > 
> > > Your patch would make sense if there was a platform with an <errno.h>
> > > that declares program_invocation_name, but without an <argp.h> file.
> > > Which platform is this?
> > 
> > uClibc exhibits this behavior, but i dont see why the platform matters. 
> > the current behavior is clearly wrong.
> 
> The platform matters, because every commit needs a rationale. You could not
> say that code that has been working fine for 8 years is "clearly wrong" if
> you couldn't exhibit a situation where it fails to work, and this
> situation is reproducible on more than one machine.

but you're relying on glibc not changing behavior.  implicit header inclusion 
is never a good idea imo.  especially considering how much work glibc has been 
doing recently to reduce header inclusion between their own headers.
-mike

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to