On Tue, 24 Jan 2017 19:01:46 +0100
Robert Varga <n...@hq.sk> wrote:
> On 01/24/2017 06:17 PM, Michael Vorburger wrote:
> >     One question around this: these look like hand-written APIs,
> > which really should be backed by a model. Is there a particular
> > reason for this?
> > 
> > using https://immutables.github.io for these could be interesting..
> > but probably more of a potential next step separate follow-up, than
> > something to mix into this. But now that there is a super generic
> > String[] API anymore, which is used to be, which this addresses,
> > moving to adopting (e.g.) Immutables.org instead of coding those
> > out as they are would be useful, IMHO.  
> 
> Actually, these things should have an associated yang model, from
> there all the stuff that immutables.org does comes out for free (in
> fact binding2 is using immutables internally).
> 
> I mean really,
> https://github.com/opendaylight/genius/blob/master/mdsalutil/mdsalutil-api/src/main/java/org/opendaylight/genius/mdsalutil/MatchInfo.java
> is tied to model-defined entity.

Yes, and it ends up building a YANG-model-defined entity.

> https://github.com/opendaylight/genius/blob/master/mdsalutil/mdsalutil-api/src/main/java/org/opendaylight/genius/mdsalutil/packet/IPv4.java
> looks exactly how AD-SAL classes used to look like...
> 
> Scratch that, it actually *is* a copy of an AD-SAL class:
> https://github.com/opendaylight/controller/blob/stable/helium/opendaylight/sal/api/src/main/java/org/opendaylight/controller/sal/packet/IPv4.java

Exactly, this code apparently dates back to the AD-SAL-to-MD-SAL
transition. I agree we should switch to the YANG builders eventually; I
started the transition mainly to get away from untyped arrays (of
String, Long or BigInteger), and it turned out it wasn't too hard to do
a clean migration for these classes. Switching completely to the YANG
models would be a lot more involved (but we should consider it!).

Thanks for your watchfulness!

Regards,

Stephen

Attachment: pgpZR6SQyNG0i.pgp
Description: OpenPGP digital signature

_______________________________________________
sfc-dev mailing list
sfc-dev@lists.opendaylight.org
https://lists.opendaylight.org/mailman/listinfo/sfc-dev

Reply via email to