Timo -

I'm a bit confused.  Am I reading your email correctly?  You agree that
zebra.h is a grab bag of unrelated information, the enum zebra_link_type
doesn't belong there and you are not going to move it?  Would you clarify
your reasoning behind not moving it?

thanks!

donald



On Wed, Jan 6, 2016 at 1:17 PM, Timo Teras <[email protected]> wrote:

> On Tue, 5 Jan 2016 10:16:28 +0000 (GMT)
> Paul Jakma <[email protected]> wrote:
>
> > On Mon, 4 Jan 2016, Donald Sharp wrote:
> >
> > > Doesn't 'enum zebra_link_type' actually belong in lib/if.h?
> > > instead of lib/zebra.h?
> >
> > No strong feelings either way, or objction from me.
>
> zebra.h seems to contain a pretty random collection of stuff. From
> fixing system headers, to most of the zclient protocol. We should
> probably move all zclient stuff to separate header that defines
> wireprotocol stuff only.
>
> But for now, I think I just add the ll_type stuff to zebra.h.
> Technically it might be desirable to use it from elsewhere; even if the
> sole user is if.[ch] now.
>
> > Two things:
> >
> > 1. ZServ is a long and ll_type an unsigned short?
>
> Fixed ll_type to be enum. Is long ok, or should it be made short?
>
> > 2. Should the (struct interface) member be the enum? In which case, a
> >     little helper to use a switch to convert between the wire type
> > and the enum might be good to ensure mismatches get caught.
>
> Hopefully the wire type is converted only in the one place. I'd rather
> not introduce helpers for the sole use case. But if that's wanted, I
> can add it. If yes, should they be just static inlines in the affected
> files, or made global lib api?
>
_______________________________________________
Quagga-dev mailing list
[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-dev

Reply via email to