)]}'
{"quantum/common/constants.py":[{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"634fd7c3ab9987faa02e27ed45459b37dbe7658a","unresolved":false,"context_lines":[{"line_number":72,"context_line":"SORT_DIRECTION_DESC \u003d \u0027desc\u0027"},{"line_number":73,"context_line":""},{"line_number":74,"context_line":"L3_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027l3_agent_scheduler\u0027"},{"line_number":75,"context_line":"DHCP_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027dhcp_agent_scheduler\u0027"}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAPn%2F%2FaPw%3D","line":75,"updated":"2013-07-02 14:50:28.000000000","message":"By splitting the alias and letting existing agents expose one or the other effectively breaks backward compatibility. I don\u0027t know how much this extension is used or relied upon by other plugins, but I wonder whether we should keep an \u0027agent_scheduler\u0027 extension that is a collection of all the agent scheduling capabilities, so far l3 and dhcp (soon lbaas). A Plugin may choose to support all of them (and hence support the \u0027agent_scheduler\u0027 alias) or just a few (and hence pick and mix).","commit_id":"1079977d4fdfd26a81f1d9079aed24ac2af9e299"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"unresolved":false,"context_lines":[{"line_number":72,"context_line":"SORT_DIRECTION_DESC \u003d \u0027desc\u0027"},{"line_number":73,"context_line":""},{"line_number":74,"context_line":"L3_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027l3_agent_scheduler\u0027"},{"line_number":75,"context_line":"DHCP_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027dhcp_agent_scheduler\u0027"}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAPn%2F%2FZ20%3D","line":75,"in_reply_to":"AAAAPn%2F%2FaF4%3D","updated":"2013-07-02 16:48:47.000000000","message":"Ok, I got your point, thanks for the clarification! Although I\u0027m not a fan of dummies and legacy code let\u0027s wait for some time to see if somebody have concerns about it.","commit_id":"1079977d4fdfd26a81f1d9079aed24ac2af9e299"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"unresolved":false,"context_lines":[{"line_number":72,"context_line":"SORT_DIRECTION_DESC \u003d \u0027desc\u0027"},{"line_number":73,"context_line":""},{"line_number":74,"context_line":"L3_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027l3_agent_scheduler\u0027"},{"line_number":75,"context_line":"DHCP_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027dhcp_agent_scheduler\u0027"}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAPn%2F%2FaF4%3D","line":75,"in_reply_to":"AAAAPn%2F%2FaLw%3D","updated":"2013-07-02 15:38:42.000000000","message":"Let me give you an example...say that an existing client wants to inspect the server capabilities by querying plugin extensions: prior to your change the plugin would advertise to support the agent_scheduler ext. After your change the extension is gone. The client (that used to work) freaks out. I am not saying this is bad and that are not ways around it, but I am wondering if that should be prevented. I am not even sure why we even advertise this extension in the first place, as this should be pretty much transparent to the user. That said, I am fine with splitting the alias in two, I was just raising the point for discussion to give plugin implementors time to chime in.","commit_id":"1079977d4fdfd26a81f1d9079aed24ac2af9e299"},{"author":{"_account_id":2874,"name":"yong sheng gong","email":"gong.yongsheng@99cloud.net","username":"gongysh"},"change_message_id":"5751e75ac3f74cad321d464437043df41542e475","unresolved":false,"context_lines":[{"line_number":72,"context_line":"SORT_DIRECTION_DESC \u003d \u0027desc\u0027"},{"line_number":73,"context_line":""},{"line_number":74,"context_line":"L3_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027l3_agent_scheduler\u0027"},{"line_number":75,"context_line":"DHCP_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027dhcp_agent_scheduler\u0027"}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAPn%2F%2FaHo%3D","line":75,"in_reply_to":"AAAAPn%2F%2FaLw%3D","updated":"2013-07-02 15:25:53.000000000","message":"sometimes, we have to trade off.  If some one insists, we can keep a dummy schedule-agent extension.","commit_id":"1079977d4fdfd26a81f1d9079aed24ac2af9e299"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"ff219a147efcb7c1ed7b9ec71ca10c89fe05e266","unresolved":false,"context_lines":[{"line_number":72,"context_line":"SORT_DIRECTION_DESC \u003d \u0027desc\u0027"},{"line_number":73,"context_line":""},{"line_number":74,"context_line":"L3_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027l3_agent_scheduler\u0027"},{"line_number":75,"context_line":"DHCP_AGENT_SCHEDULER_EXT_ALIAS \u003d \u0027dhcp_agent_scheduler\u0027"}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAPn%2F%2FaLw%3D","line":75,"in_reply_to":"AAAAPn%2F%2FaPw%3D","updated":"2013-07-02 15:07:30.000000000","message":"Sorry, I\u0027m not completely sure what backward compatibility do you mean. All plugins that supported \u0027agent_scheduler\u0027 extension now support two separated extensions and all functionality remains the same. In future when new scheduling extensions are arrived each plugin may start supporting them by simple adding extension alias to its supported list and deriving corresponding mixin class. Could you please clarify the benefits of common \u0027agent_scheduler\u0027 alias?","commit_id":"1079977d4fdfd26a81f1d9079aed24ac2af9e299"}]}