On Dec 11, 2012, at 9:55 PM, Rob Sherwood <rob.sherw...@bigswitch.com> wrote:
> Is it your interpretation that for OF1.0, the spec is *clear* that one
> should not match on cookie in a DELETE or do you think it is ambiguous
> and you've chosen the interpretation that it does not? More to the
> point: if I were to submit a patch to change the behavior to match on
> cookie in this situation, would you be inclined to accept it?
My recollection of those meetings was that having the ability to delete and
modify was explicitly forbidden. It was a hard fought fight to get cookies in
the first place, and one of the compromises those of us who wanted them made
was that controllers were explicitly not allowed to match on cookie. Here's
the message where I laid out an argument for them and how they could be
implemented (note that only OFPFC_ADD is mentioned):
https://mailman.stanford.edu/pipermail/openflow-spec/2009-October/000573.html
If you look a bit earlier for cookies in the archive, you can see some of the
discussion around them (although much of it was hashed out in the weekly
in-person meetings.)
I agree that it was silly to not allow matching, but I felt it was more
important to just have them in, since we could add the ability to match in the
next revision of OpenFlow. Speaking of...
> Ali: for my part, I think it makes sense to include it and that's how
> I interprete the spec. Certainly from a functionality stand point,
> you can imagine wanting to delete by cookie in OF1.0 (IMHO, this is
> what David Erickson -- the feature's creator -- originally intended).
> All OF1.1 added was the ability to match on a subset of the cookie.
I believe David's proposal was for 1.1. Here it is:
https://mailman.stanford.edu/pipermail/openflow-spec/2010-August/001077.html
That message was sent in August 2010, but OpenFlow 1.0.0 was released in
December 2009.
--Justin
_______________________________________________
openflow-discuss mailing list
openflow-discuss@lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
