)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":13734,"name":"Nell Jerram","email":"nell@tigera.io","username":"neiljerram"},"change_message_id":"72757d09b7eb76c6b18b51398a0ad234aef8de0c","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Remove VIF_TYPES constant"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"VIF_TYPES[1] lists all vif types including out-of-tree ones BUT is only"},{"line_number":10,"context_line":"used by out-of-tree bigswitch code. Dependant change defines VIF_TYPES"},{"line_number":11,"context_line":"in bigswitch repo so this change can remove VIF_TYPES from neutron"},{"line_number":12,"context_line":"code."},{"line_number":13,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":5,"id":"fa1b9901_63e6a916","line":10,"updated":"2015-08-20 09:02:57.000000000","message":"s/Dependant/Dependent/\n\nHowever, is it a good idea to describe this dependency relationship in the commit message here?  When these changes are merged, I presume that the move-to-bigswitch one will be merged first (as it is dependent), and this one after that.  In that context, this commit message will read wrongly, because the \"only used by out-of-tree bigswitch code\" will not be true at the point when this change is merged.","commit_id":"82245087f57b0140cc27453cef0d2453b8bf0bb4"},{"author":{"_account_id":13734,"name":"Nell Jerram","email":"nell@tigera.io","username":"neiljerram"},"change_message_id":"f2b12d71a1ffeb5be081652f106427f17c1b5435","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Remove VIF_TYPES constant"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"VIF_TYPES[1] lists all vif types including out-of-tree ones BUT is only"},{"line_number":10,"context_line":"used by out-of-tree bigswitch code. Dependant change defines VIF_TYPES"},{"line_number":11,"context_line":"in bigswitch repo so this change can remove VIF_TYPES from neutron"},{"line_number":12,"context_line":"code."},{"line_number":13,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":5,"id":"fa1b9901_837f3d16","line":10,"in_reply_to":"fa1b9901_63e6a916","updated":"2015-08-20 09:05:20.000000000","message":"I see that I got the dependency order the wrong way round!\n\nBut I think it would be clearer if it was in the order that I expected - i.e. first move the ones that are needed out-of-tree.  Then remove everything that can be removed from core.","commit_id":"82245087f57b0140cc27453cef0d2453b8bf0bb4"}],"doc/source/devref/contribute.rst":[{"author":{"_account_id":6524,"name":"Henry Gessau","email":"HenryG@gessau.net","username":"gessau"},"change_message_id":"cdeea8d7f82d585328dc8996574a6c3f7b88f995","unresolved":false,"context_lines":[{"line_number":521,"context_line":""},{"line_number":522,"context_line":"**ToDo: Interface Driver port bindings.**"},{"line_number":523,"context_line":"    These are currently defined by the ``VIF_TYPES`` in"},{"line_number":524,"context_line":"    ``neutron/extensions/portbindings.py``. We could make this config-driven"},{"line_number":525,"context_line":"    for agents. For Nova, selecting the VIF driver can be done outside of"},{"line_number":526,"context_line":"    Neutron (using the new `os-vif python library"},{"line_number":527,"context_line":"    \u003chttps://review.openstack.org/193668\u003e`_?). Armando and Akihiro to discuss."}],"source_content_type":"text/x-rst","patch_set":2,"id":"fa1b9901_d64b352a","side":"PARENT","line":524,"updated":"2015-08-18 04:45:21.000000000","message":"The goal changes slightly, but does not go away.\n\n\"The ``VIF_TYPE_*`` constants in ``neutron/extensions/portbindings.py`` should be moved from neutron core to the repositories where their drivers are implemented. We need to provide some config or hook mechanism for VIF types to be registered by external interface drivers. For Nova ...\"","commit_id":"bf800f9c6cec7ca012b8f16ea93e4bb1453017a2"},{"author":{"_account_id":6524,"name":"Henry Gessau","email":"HenryG@gessau.net","username":"gessau"},"change_message_id":"50cb7edb17e364494f97384f3b3eadd0f2e9fd49","unresolved":false,"context_lines":[{"line_number":521,"context_line":""},{"line_number":522,"context_line":"**ToDo: Interface Driver port bindings.**"},{"line_number":523,"context_line":"    These are currently defined by the ``VIF_TYPES`` in"},{"line_number":524,"context_line":"    ``neutron/extensions/portbindings.py``. We could make this config-driven"},{"line_number":525,"context_line":"    for agents. For Nova, selecting the VIF driver can be done outside of"},{"line_number":526,"context_line":"    Neutron (using the new `os-vif python library"},{"line_number":527,"context_line":"    \u003chttps://review.openstack.org/193668\u003e`_?). Armando and Akihiro to discuss."}],"source_content_type":"text/x-rst","patch_set":2,"id":"fa1b9901_d06c879e","side":"PARENT","line":524,"in_reply_to":"fa1b9901_958fe122","updated":"2015-08-18 12:58:52.000000000","message":"Admittedly I do not have enough knowledge of the problem space to give a final answer. My understanding is, and please correct me if I am wrong:\n\n- There are different VIF types.\n\n- Each VIF type needs an interface driver.\n\n- The interface drivers can be in-tree or external.\n\n- VIF_TYPE_X could have one driver in repo-A and another in repo-B, but only one driver can be used in a deployment, so repo-A and repo-B need to use the same VIF type name to detect conflict.\n\n- If repo-P and repo-Q each implement an interface driver for different VIF types, we need to ensure that they use different VIF type names.","commit_id":"bf800f9c6cec7ca012b8f16ea93e4bb1453017a2"},{"author":{"_account_id":8124,"name":"cbrandily","email":"zzelle@gmail.com","username":"cbrandily"},"change_message_id":"bd3a52b7ea15cc4ba05979e231ded5ebbfe9a5f9","unresolved":false,"context_lines":[{"line_number":521,"context_line":""},{"line_number":522,"context_line":"**ToDo: Interface Driver port bindings.**"},{"line_number":523,"context_line":"    These are currently defined by the ``VIF_TYPES`` in"},{"line_number":524,"context_line":"    ``neutron/extensions/portbindings.py``. We could make this config-driven"},{"line_number":525,"context_line":"    for agents. For Nova, selecting the VIF driver can be done outside of"},{"line_number":526,"context_line":"    Neutron (using the new `os-vif python library"},{"line_number":527,"context_line":"    \u003chttps://review.openstack.org/193668\u003e`_?). Armando and Akihiro to discuss."}],"source_content_type":"text/x-rst","patch_set":2,"id":"fa1b9901_958fe122","side":"PARENT","line":524,"in_reply_to":"fa1b9901_d64b352a","updated":"2015-08-18 12:41:49.000000000","message":"\"The goal changes slightly, but does not go away.\" \n\nNot sure to understand, neutron doesn\u0027t use VIF_TYPES and according to git log -SVIF_TYPES, bigswitch plugin was/is the only one to use VIF_TYPES in neutron. So such vif types registering mechanism[1] should be provided and documented in networking-bigswitch?\n\n[1] a simple solution would be to define a new option to declare out-of-tree vif types innetworking-bigswitch repo?","commit_id":"bf800f9c6cec7ca012b8f16ea93e4bb1453017a2"}]}