Comment 11 for bug 1182384

This was the conversation I had with Sean Dague in IRC back on 6/18:

http://paste.openstack.org/show/44543/

When Sean is talking about catching the 400 and translating it to a 404, I think that's tied to the second option in the mailing list thread here:

http://lists.openstack.org/pipermail/openstack-dev/2013-July/011666.html

"> 2) Encapsulate all calls to validate_id() in a try/catch HTTPBadRequest and raise a HTTPNotFound instead (exception translation)"

However, later when I talked to Sean on 7/24 he suggested I could try to fix the nova security group id validation issue by using this patch as a template:

https://review.openstack.org/#/c/34559/

In that patch nova-network and neutron are validating name/description differently (neutron wasn't actually validating name/description), and it tries to move the validation into the base class. I'm not actually sure how that applies to the id validation since that's implemented in both the nova-network and neutron SecurityGroupBase implementations in nova.

So it sounds like Sean thinks we should return 404 in either case when the id is invalid, but I've also seen the proposed tempest patch to fix the nova tests in tempest if neutron is being used, and I didn't understand the problem with those (maybe because the client is forced to know the api implementation in nova?).

Anyway, could probably follow up on the mailing list or try to get a discussion going in #openstack-nova IRC?