Peter Memishian wrote:
> So why don't we just remove that capability from ifconfig, then?
We'd risk backward compatibility, plus I don't think there's anything
wrong with using the subcommand if the interface is not in-use (e.g., no
IFF_UP addresses).
There's nothing wrong with SIOCSLIFINDEX being used as long as nothing
is actually using the index in any meaningful manner. If there are
conditions in which it is dangerous and safe, we need to document those,
along with what's required to play safe when it comes to using the
interface index.
If we're not prepared to document those cases then we should do
something else. I don't see it as being acceptable to have a
component of Solaris networking that doesn't play well with
others, either by design or by accident, especially when we hang
it out there for people to trip over.
I cannot figure out why this is such a concern. Did a customer report a
problem?
The original problem (stuffing the index into b_prev) is the concern.
To me, this is puke material (but then so is much of the other abuse
of this field) and after having seen it, I was hoping that something
could be done to clean it up. Until I knew about it, it wasn't a
problem but now I do. I'm not really happy with the situation and
it won't go away if I just close my eyes and pretend it isn't there.
Maybe I'll just come up with something, anyway, to fix this bug and
do the usual dance with that.
Cheers,
Darren
_______________________________________________
networking-discuss mailing list
[email protected]
