On Fri, Sep 04, 2009 at 05:44:52PM +0200, Marc Lehmann wrote:
> On Fri, Sep 04, 2009 at 05:31:23PM +0200, Marc Lehmann <[email protected]> 
> wrote:
> > > includes functions beyond SUS. I don't know what Roy wrote in his
> > > original message, I am just telling you that blaming NetBSD for adding
> > > common functions to the default namespace is off.
> 
> Funny how you yourself (or another Joerg Sonnenberger)
> called it a bug to provide them by default in string.h in this thread:
> http://archive.netbsd.se/?ml=netbsd-tech-userlevel&a=2009-08&t=11235091
> 
> Also funny how you call this gnu compatibility when the gnu system itself
> uses the implementation namespace and avoids clashes. Likewise for gcc,
> which calls it __builtin_popcount for a reason.
> 
> But yeah, it's off to blame netbsd for being the only one to not do it
> sanely.

I will ignore the rest as it is obvious that you don't even bother to
check your facts. The BSD extensions are enabled in glibc as well by
default, just like on the BSDs. The only major difference is that
NetBSD provides ffs and related functions in strings.h, which is a XSI
extension and conditionally includes that one while glibc pulls them in
directly in string.h if the conditionals are not set.

GCC is a completely different issue. As compiler it is not allowed to
add to the namespace without the __ protection. It often enough
implements that rule gracefully when not in standalone mode. To the
point that many of the GCC builtins are known to be slower than the libc
versions at least on NetBSD for the non-trivial cases.

In short, NetBSD is as sane as glibc in this regard. The rest is up to
the reader to decide...

Joerg

_______________________________________________
rxvt-unicode mailing list
[email protected]
http://lists.schmorp.de/cgi-bin/mailman/listinfo/rxvt-unicode

Reply via email to