Re: gdomap bug - patch

[Richard Frith-Macdonald]

On Friday, January 25, 2002, at 10:19 PM, Pete French wrote:

> > Thanks ... I added your patch with a slight modification and all the
> > associated autoconf stuff to decide whether the system has variable
>
> Hmm, how did you do that ? It doesnt seem to work here (and I re-ran
> the configure script). Is it supposed to set HAVE_SA_LEN if the
> system has variable length records ? It doesnt seem to do so
> on this machine...
>
> I dont know of a way to detect this behaviour either, how is
> autoconf doing it ? I did think about tyring to detect this at runtime,
> but couldnt come up with a suitable way to do it. A Linux box should
> have the sa_len field in it anyway, and hopefully it will be correct?

See configure.in ... it tries to compile a tiny test program.  Perhaps
it needs other includes rather than net/if.h on BSD?

> > data structures with an sa_length field or not ... presumably only the
> > recent BSDs do (certainly linux doesn't).
>
> I was assuming it would work unmodified on older systems. What
> does Linux give for the contents of sa_len in the socket structure ?

The sa_len field does not exist at all on any of my systems.

> The only older UNIX I have to hand is the OpenStep BSD 4.3 system
> and I havent had a chance to try that yet.
>
> BTW, I did notice that I made an error in my code anyway - the pointer
> advance line should read:
>
> ifr_ptr += sizeof(ifreq) - sizeof(ifreq.ifr_addr) + 
> ifreq.ifr_addr.sa_len;

fixed that.