Re: ifconfig and some minor fixes.

[Marcus Brinkmann]

On Wed, Jun 20, 2001 at 11:42:18AM -0400, Alain Magloire wrote:
> > 
> > On Wed, Jun 20, 2001 at 12:14:47AM -0400, Alain Magloire wrote:
> > > 
> > > QNX/Neutrino does not provide rfc2133 if_nameindex() suite, unfortunate.
> > > Can we implement something and autodetect if_nameindex() and suite.
> > 
> > This is what we have to do, plus an override for systems where if_nameindex
> > is there but broken.  I hope to come back to this later this week (the
> > changes are still pending).
> 
> Not a problem.  I can provide an implemetation if you wish.
> Actually I think there is one part of glibc(sorry don't remember
> the file ... wait ... let me check ....)
> ....
> Found it !  glibc-2.2.2/sysdeps/unix/sysv/linux/if_index.c
> 
> This one should do fine after kosher to remove the glibc bits.

That was exactly what I was intending to do.  The existing bits in the
file are actually cut and pasted from there.

If you want to fix it up, be my guest.

> - oops! Just Noticed if_index.c, SA_LEN() should not be use
> It was part of an older draft and use on BSD.  Later it was
> pull out of the draft.  You should not rely on it to exist.
> And if fact it does not exist on QNX/Neutrino.

Thanks.  I am actually not very fluent in this realm.  Can you fix it properly?
(I will then pick up your patch and try to learn it for the future).

> - sa_family_t  does not exist on Neutrino, can we provide
> a check for it in configure.in  and define in config.h
>       typedef unsigned short int sa_family_t;
> or in the appropriate file.

I think Sergej (sp?)  already submitted patches for that.

> - ifr->ifr_netmask ??? Could not find a definition
> for this ... but SIOCGIFNETMASK was defined.
> I did not investigate, running out of cycles.

This is strange.  You should probably quote from the header what is there.
Maybe they expect us to use ifr_addr (see the soalris fuile for a similar hack).

Marcus