On Thu, Jan 10, 2013 at 12:55:09PM -0800, Ethan Jackson wrote:
> Out of curiosity, did you consider the name xassert() as an alternative to
> ovs_assert()? It has the advantage of following our established convention.
I was taking the "x" prefix to indicate a function wrapper that aborts
on error instead of reporting the error. Since assert() already
aborts on error, it doesn't really follow the pattern.
> + static int reentry;
> >
>
> I'm not sure if the C spec strictly requires it, but I would feel more
> comfortable if we initialized this to 0 explicitly.
OK, I'll do that. (It isn't strictly required.)
> > + switch (reentry++) {
> > + case 0:
> > + VLOG_ABORT("%s: assertion %s failed in %s()",
> > + where, condition, function);
> > + /* not reached */
> >
> Minor style bit I may go ahead and make this an abort() for clarity. I.E.
> ovs_abort(); /* not reached */
I'd be more comfortable with a plain abort(), if that's OK.
> Acked-by: Ethan Jackson <et...@nicira.com>
Thanks.
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
