)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":17444,"name":"Jay.Shen","email":"shenhong09@huawei.com","username":"Jay.Shen"},"change_message_id":"064b4e5850d8efb9cf336cac071bbb40203905e2","unresolved":false,"context_lines":[{"line_number":12,"context_line":""},{"line_number":13,"context_line":"Note there are two proposals in the spec now, a"},{"line_number":14,"context_line":"compromise \"Option B\". I\u0027d be interested in folks"},{"line_number":15,"context_line":"puting some thought into this and providing feedback."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"This does not apply to the current reference implementation"},{"line_number":18,"context_line":"LVM iSCSI driver."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"1a4dcd0f_baec4fc8","line":15,"updated":"2015-08-04 11:02:52.000000000","message":"S/puting/putting","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"}],"specs/liberty/make-plugins-external-libs.rst":[{"author":{"_account_id":12924,"name":"Patrick East","email":"east.patrick@gmail.com","username":"patrick.east"},"change_message_id":"c3ffaccd932998b1d26e85b86359e3309fe6c46b","unresolved":false,"context_lines":[{"line_number":11,"context_line":"Cinder has always offered a plugin model to provide a choice of"},{"line_number":12,"context_line":"backend devices/drivers.  This has been great, for a number of things:"},{"line_number":13,"context_line":"1. Bringing sponsorship to the Cinder project in the form of vendor support"},{"line_number":14,"context_line":"2. Validating the Cinder service in terms of giving \"customers\" a warm fuzzy"},{"line_number":15,"context_line":"3. Offering choice for people that have varying resources available to them"},{"line_number":16,"context_line":"   and want to deploy an OpenStack Cloud"},{"line_number":17,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"7a016987_dba1fbaa","line":14,"updated":"2015-05-28 01:06:19.000000000","message":"a warm fuzzy what? :p","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":2243,"name":"John Griffith","email":"john.griffith8@gmail.com","username":"john-griffith"},"change_message_id":"8ccd3197808f8bbb1268714b7af1a0a7db601169","unresolved":false,"context_lines":[{"line_number":11,"context_line":"Cinder has always offered a plugin model to provide a choice of"},{"line_number":12,"context_line":"backend devices/drivers.  This has been great, for a number of things:"},{"line_number":13,"context_line":"1. Bringing sponsorship to the Cinder project in the form of vendor support"},{"line_number":14,"context_line":"2. Validating the Cinder service in terms of giving \"customers\" a warm fuzzy"},{"line_number":15,"context_line":"3. Offering choice for people that have varying resources available to them"},{"line_number":16,"context_line":"   and want to deploy an OpenStack Cloud"},{"line_number":17,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"5afe65bd_f53f6dd4","line":14,"in_reply_to":"7a016987_dba1fbaa","updated":"2015-06-02 07:12:05.000000000","message":"LOL\n\u003cinsert-favorite-warm-fuzzy-object-here\u003e (default\u003d\"feeling\")","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"855c383794f5fdedb3ec6083e7aff72faec53b26","unresolved":false,"context_lines":[{"line_number":35,"context_line":"the effectiveness of the project itself."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Once the number of LOC for plugins approaches the LOC of the project itself"},{"line_number":38,"context_line":"it starts one should start to wonder what the project actually provides."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_6ea047f5","line":38,"updated":"2015-06-05 14:55:32.000000000","message":"This does not appear to describe an actual problem.  It looks like it\u0027s just an opinion?","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"855c383794f5fdedb3ec6083e7aff72faec53b26","unresolved":false,"context_lines":[{"line_number":41,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"Removing the plugins from treee should result in the development focus of Cinder"},{"line_number":44,"context_line":"to actually be \"the development of Cinder\" as it was originally intended."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"Proposed change"},{"line_number":47,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_ee48b72f","line":44,"updated":"2015-06-05 14:55:32.000000000","message":"I know this is the hope, but I\u0027m not sure it\u0027s that easy to predict that fewer drivers to look at \u003d more productive development on core parts of Cinder.","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"855c383794f5fdedb3ec6083e7aff72faec53b26","unresolved":false,"context_lines":[{"line_number":65,"context_line":"Third party CI would still be a requirements, and in order for the vendors"},{"line_number":66,"context_line":"plugin to be able to be configured in Cinder it would be required that the"},{"line_number":67,"context_line":"driver is still \"approved\" however the review approval process would now"},{"line_number":68,"context_line":"simply require a working and stable CI system."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"This would also hopefully bring some focus to the reference implementation"},{"line_number":71,"context_line":"(LVM) driver."}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_4e090b2c","line":68,"updated":"2015-06-05 14:55:32.000000000","message":"This needs a lot more definition.  Really, I\u0027m not sure how this is even possible if I\u0027m understanding the proposal.\n\nHow are we going to prevent a driver from being configured in Cinder?\n\nToday people can load out-of-tree drivers like other drivers by putting them in the right path etc... this would no longer be possible?\n\nI don\u0027t really see how this is feasible or makes sense in an open source platform.","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"855c383794f5fdedb3ec6083e7aff72faec53b26","unresolved":false,"context_lines":[{"line_number":68,"context_line":"simply require a working and stable CI system."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"This would also hopefully bring some focus to the reference implementation"},{"line_number":71,"context_line":"(LVM) driver."},{"line_number":72,"context_line":""},{"line_number":73,"context_line":"Alternatives"},{"line_number":74,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_ae175fec","line":71,"updated":"2015-06-05 14:55:32.000000000","message":"Again, this is a pretty big assumption that significantly more people will become interested in the LVM driver...","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"855c383794f5fdedb3ec6083e7aff72faec53b26","unresolved":false,"context_lines":[{"line_number":91,"context_line":""},{"line_number":92,"context_line":"Security impact"},{"line_number":93,"context_line":"---------------"},{"line_number":94,"context_line":""},{"line_number":95,"context_line":"None"},{"line_number":96,"context_line":""},{"line_number":97,"context_line":"Notifications impact"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_aee4ff3a","line":94,"updated":"2015-06-05 14:55:32.000000000","message":"Drivers will no longer have nearly as much review for security problems.  Security problems aren\u0027t going to be found by running CI.","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"855c383794f5fdedb3ec6083e7aff72faec53b26","unresolved":false,"context_lines":[{"line_number":101,"context_line":""},{"line_number":102,"context_line":"Other end user impact"},{"line_number":103,"context_line":"---------------------"},{"line_number":104,"context_line":""},{"line_number":105,"context_line":"None"},{"line_number":106,"context_line":""},{"line_number":107,"context_line":"Performance Impact"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_aeee3fea","line":104,"updated":"2015-06-05 14:55:32.000000000","message":"Increased corner-case problems with drivers, due to lower code quality since they are no longer being reviewed nearly as well.","commit_id":"d3e68add899134bab1a8d1ff81fab72f436fe1ce"},{"author":{"_account_id":16708,"name":"Kendall Nelson","display_name":"Kendall (diablo_rojo)","email":"kennelson11@gmail.com","username":"kjnelson"},"change_message_id":"205fb268d8147edc3a7b85a7c8e97434cc1fe13f","unresolved":false,"context_lines":[{"line_number":35,"context_line":"from the effectiveness of the project itself only because how many there are now."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Once the number of LOC for plugins approaches the LOC of the project itself"},{"line_number":38,"context_line":"one should start to wonder what the project actually provides."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":2,"id":"fa1b9901_e8c1e2e4","line":38,"updated":"2015-08-21 15:56:23.000000000","message":"I agree with Tom. I think this argument would benefit from more specific numbers about growth rate, LOC\u0027s, etc.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":9003,"name":"Tom Barron","email":"tpb@dyncloud.net","username":"tbarron"},"change_message_id":"8c617f76eb83cba16c381bb224d91d84ab27343f","unresolved":false,"context_lines":[{"line_number":35,"context_line":"from the effectiveness of the project itself only because how many there are now."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Once the number of LOC for plugins approaches the LOC of the project itself"},{"line_number":38,"context_line":"one should start to wonder what the project actually provides."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_4f9a2216","line":38,"updated":"2015-06-05 17:47:09.000000000","message":"Maybe state explicitly that the growth rate of driver code and patches, and the corresponding need to review these, drains resources that could instead be dedicated to core maintenance and development.  Be explicit about what the growth rate is, that it is accelerating, that it has no apparent bound (?), etc.\n\nAnd maybe state explicitly that stackalytics and LOC counts may create incentives for unproductive code churn in vendor code that exacerbates the first problem.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":7350,"name":"Jordan Pittier","email":"jordan.pittier@gmail.com","username":"JordanP"},"change_message_id":"ba516f88485fe7cf0a1f8a4f776c220d1a02f87c","unresolved":false,"context_lines":[{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Once the number of LOC for plugins approaches the LOC of the project itself"},{"line_number":38,"context_line":"one should start to wonder what the project actually provides."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":42,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba3cc151_33305001","line":39,"updated":"2015-06-30 15:51:06.000000000","message":"Another problem is that, as a vendor, it takes weeks before my small change to my driver gets in.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":17444,"name":"Jay.Shen","email":"shenhong09@huawei.com","username":"Jay.Shen"},"change_message_id":"064b4e5850d8efb9cf336cac071bbb40203905e2","unresolved":false,"context_lines":[{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"Removing the plugins from treee should result in the development focus of Cinder"},{"line_number":44,"context_line":"to actually be \"the development of Cinder\" as it was originally intended."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"Proposed change"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1a4dcd0f_ba40cf57","line":43,"updated":"2015-08-04 11:02:52.000000000","message":"S/treee/tree","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":16708,"name":"Kendall Nelson","display_name":"Kendall (diablo_rojo)","email":"kennelson11@gmail.com","username":"kjnelson"},"change_message_id":"205fb268d8147edc3a7b85a7c8e97434cc1fe13f","unresolved":false,"context_lines":[{"line_number":57,"context_line":""},{"line_number":58,"context_line":"Everybody is afraid to attempt this because the company they"},{"line_number":59,"context_line":"represent won\u0027t be able to obtain bragging rights for what they commit,"},{"line_number":60,"context_line":"how many drivers they have in tree etc."},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"This spec would propose that we remove the drivers from tree and that"},{"line_number":63,"context_line":"each vendor/plugin other than the reference implementation would provide"}],"source_content_type":"text/x-rst","patch_set":2,"id":"fa1b9901_5d120f5a","line":60,"updated":"2015-08-21 15:56:23.000000000","message":"As unfortunate a truth as it is, I think that Duncan is right with his statement that vendors need to be incentivized to care and that the threat of being taken out of tree is enough to get people to act at this point. If that ceases to be the case, and people don\u0027t care about being out of tree then we could take the suggested approach here and remove them.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":9003,"name":"Tom Barron","email":"tpb@dyncloud.net","username":"tbarron"},"change_message_id":"8c617f76eb83cba16c381bb224d91d84ab27343f","unresolved":false,"context_lines":[{"line_number":57,"context_line":""},{"line_number":58,"context_line":"Everybody is afraid to attempt this because the company they"},{"line_number":59,"context_line":"represent won\u0027t be able to obtain bragging rights for what they commit,"},{"line_number":60,"context_line":"how many drivers they have in tree etc."},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"This spec would propose that we remove the drivers from tree and that"},{"line_number":63,"context_line":"each vendor/plugin other than the reference implementation would provide"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_e541ed3a","line":60,"updated":"2015-06-05 17:47:09.000000000","message":"I don\u0027t think that this is a big source of resistance to the \"out of tree\" proposal by *this* group, though I know there is a dynamic where companies count lines of code and brag.\n\nHave we considered working with the stackalytics folks to only have core LOC reported?  It seems to me that that could be done easily enough by putting vendor driver code under known subdirectory trees (like exclude these files from backup).\n\nThat by itself could help with the \"motivate companies to work on core\" issue and stop companies from getting bragging rights just because they refactor every release.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":7350,"name":"Jordan Pittier","email":"jordan.pittier@gmail.com","username":"JordanP"},"change_message_id":"ba516f88485fe7cf0a1f8a4f776c220d1a02f87c","unresolved":false,"context_lines":[{"line_number":57,"context_line":""},{"line_number":58,"context_line":"Everybody is afraid to attempt this because the company they"},{"line_number":59,"context_line":"represent won\u0027t be able to obtain bragging rights for what they commit,"},{"line_number":60,"context_line":"how many drivers they have in tree etc."},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"This spec would propose that we remove the drivers from tree and that"},{"line_number":63,"context_line":"each vendor/plugin other than the reference implementation would provide"}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba3cc151_13998c1f","line":60,"in_reply_to":"3afb71cf_e541ed3a","updated":"2015-06-30 15:51:06.000000000","message":"I think some companies (mine for example) are afraid to have an out of tree driver because, in meetings with customers, there\u0027s always the question \u0027is your driver integrated ?\u0027 And if you reply \"no\", it\u0027s bad for you as a vendor.\n\nNow if everybody is out, then this changes.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":1207,"name":"Duncan Thomas","email":"duncan.thomas@gmail.com","username":"duncan-thomas"},"change_message_id":"11e34c0b3dd7eea92b4d0272432d77e97ae70ac6","unresolved":false,"context_lines":[{"line_number":57,"context_line":""},{"line_number":58,"context_line":"Everybody is afraid to attempt this because the company they"},{"line_number":59,"context_line":"represent won\u0027t be able to obtain bragging rights for what they commit,"},{"line_number":60,"context_line":"how many drivers they have in tree etc."},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"This spec would propose that we remove the drivers from tree and that"},{"line_number":63,"context_line":"each vendor/plugin other than the reference implementation would provide"}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba3cc151_21421311","line":60,"in_reply_to":"ba3cc151_13998c1f","updated":"2015-07-01 09:26:40.000000000","message":"The fact that out-tree drivers are seen as less good than in-tree drivers is exactly what I *like* about the current system. It gives cinder core leverage on driver developers (\u0027We\u0027ll remove your driver from the tree\u0027 is a significant threat). I think if we lose that, then the quality of drivers will go way down, rapidly. It is an absolute fact that half the drivers in the tree plain didn\u0027t work (for varieties of definitions of \u0027work\u0027, from \u0027loads at all\u0027 to \u0027all cinder operations work correctly\u0027) until we forced the issue with CI. Quite simply, vendors need to be incentivised to care. If you care about cinder as a solution, then we need good drivers.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":9003,"name":"Tom Barron","email":"tpb@dyncloud.net","username":"tbarron"},"change_message_id":"8c617f76eb83cba16c381bb224d91d84ab27343f","unresolved":false,"context_lines":[{"line_number":66,"context_line":""},{"line_number":67,"context_line":"Third party CI would still be a requirements, and in order for the vendors"},{"line_number":68,"context_line":"plugin to be able to be configured in Cinder it would be required that the"},{"line_number":69,"context_line":"driver is still \"approved\" however the review approval process would now"},{"line_number":70,"context_line":"simply require a working and stable CI system."},{"line_number":71,"context_line":""},{"line_number":72,"context_line":"This would also hopefully bring some focus to the reference implementation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_b21a1916","line":69,"updated":"2015-06-05 17:47:09.000000000","message":"It might make sense to have a list of \"certified\" drivers in the core code, with signatures or something, so that when drivers plug in they register as certified (and log, notify, etc. accordingly).   No attempt to make un-certified (e.g. development, experimental, roll-your own) drivers break, just informational.\n\nJust running CI reliably and consistently is still a challenge for many drivers, but that would seem to be a minimal requirement for certification.\n\nNot causing performance problems, corrupting the database, introducing security problems, and other things I haven\u0027t thought of should likely also be required for certification.\n\nA big question is whether when you add everything in that should be there for certification, do you have a system/process that requires less resources than (1) the current system, or (2) a reform of the current system that keeps driver code in the same git repo.\n\nCould we spend lot less time reviewing vendor-specific code if we had a different set of guidelines than for core code?  For example:\n\n* We care if you give us code to review that is painful to read and is therefore a waste of my time \u0026 we will down vote it for that without continuing further in the review.\n\n* We down-vote for obvious gaffes and bad external effects:  api issues, bad database interaction, performance degradation, security problems, ...\n\n* We do not down-vote for spelling, minor coding style preferences, unit test coverage issues, logging issues that can\u0027t be detected by automated hacking tests, etc.  It\u0027s your code and if you aren\u0027t harming the rest of the project then weakness in these areas just reflects on you and your company.  Don\u0027t depend on community reviews to get the quality that you should get from internal review before putting your code up for community review.\n\nIf we did something like the above it might make sense to use a directory structure something like the following:\n\n   cinder\n      - volume\n            -- drivers\n                  --- lvm\n                  --- vendor\n                       ---- vendor X\n                       ---- vendor Y\n                      ...\n   and similarly under cinder/tests/unit/..., etc.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":1736,"name":"Ivan Kolodyazhny","email":"e0ne@e0ne.info","username":"e0ne"},"change_message_id":"595252f3a644a56d8ba53bdb3dc2fa0f851c6fa9","unresolved":false,"context_lines":[{"line_number":66,"context_line":""},{"line_number":67,"context_line":"Third party CI would still be a requirements, and in order for the vendors"},{"line_number":68,"context_line":"plugin to be able to be configured in Cinder it would be required that the"},{"line_number":69,"context_line":"driver is still \"approved\" however the review approval process would now"},{"line_number":70,"context_line":"simply require a working and stable CI system."},{"line_number":71,"context_line":""},{"line_number":72,"context_line":"This would also hopefully bring some focus to the reference implementation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1a4dcd0f_82ed5cf1","line":69,"in_reply_to":"3afb71cf_b21a1916","updated":"2015-08-04 14:50:19.000000000","message":"Getting a list of a certified out-of-tree drivers is a good idea. But it should be simple enough for Cinder maintainers.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":12924,"name":"Patrick East","email":"east.patrick@gmail.com","username":"patrick.east"},"change_message_id":"b80ef14c3340c5fb9d724a08c7d62701fd7ea002","unresolved":false,"context_lines":[{"line_number":78,"context_line":"tricky in Git and it introduces some complexities in how we test and ensure"},{"line_number":79,"context_line":"compatibility going forward."},{"line_number":80,"context_line":""},{"line_number":81,"context_line":"I think to start we certainly would want to avoid the subrepo idea, but I"},{"line_number":82,"context_line":"did want to suggest it as a possibility."},{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Alternatives"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_bd00afab","line":81,"updated":"2015-06-05 16:43:16.000000000","message":"+1 subrepo\u0027s are bad news","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":1207,"name":"Duncan Thomas","email":"duncan.thomas@gmail.com","username":"duncan-thomas"},"change_message_id":"a44170661081db7e2657a8a7cdf6db474ab96bbc","unresolved":false,"context_lines":[{"line_number":106,"context_line":"    hesitation I have is that if you still have the general rules of +1 and multiple"},{"line_number":107,"context_line":"    +2s from core members it solve absolutely nothing.  If on the other hand there"},{"line_number":108,"context_line":"    is a different process instituted, call it \"relaxed\", \"special\" or whatever you"},{"line_number":109,"context_line":"    like; then I think it is something that might be incrementally better."},{"line_number":110,"context_line":""},{"line_number":111,"context_line":"    One thing that I would propose if we went this route would be that we set it up"},{"line_number":112,"context_line":"    with an independent release cycle.  So manage it similar to cinderclient in terms"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_28411c0c","line":109,"updated":"2015-06-05 18:28:58.000000000","message":"I think we can have more people with +2 abilities on that repo, and a honour system of \u0027don\u0027t +2 your own / your company\u0027s drivers\u0027. If driver authors won\u0027t review each other\u0027s stuff then they make slow progress - that\u0027s something they can, to a degree, sort between themselves.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":9003,"name":"Tom Barron","email":"tpb@dyncloud.net","username":"tbarron"},"change_message_id":"8c617f76eb83cba16c381bb224d91d84ab27343f","unresolved":false,"context_lines":[{"line_number":113,"context_line":"    of updating it and pushing changes out.  This will require some effort obviously"},{"line_number":114,"context_line":"    to deal with compatibility etc, but I think it\u0027s worth it, and I think that a"},{"line_number":115,"context_line":"    rolling driver model might help alleviate some of the problems I see with"},{"line_number":116,"context_line":"    Cinder development."},{"line_number":117,"context_line":""},{"line_number":118,"context_line":"Data model impact"},{"line_number":119,"context_line":"-----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_ef5e6e44","line":116,"updated":"2015-06-05 17:47:09.000000000","message":"3. Lieutenant system with the current repo ?","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":5997,"name":"Walt","display_name":"Hemna","email":"waboring@hemna.com","username":"walter-boring","status":"SAP"},"change_message_id":"f3f703b0ca845d34bbc88aef843dcdcddaad8110","unresolved":false,"context_lines":[{"line_number":113,"context_line":"    of updating it and pushing changes out.  This will require some effort obviously"},{"line_number":114,"context_line":"    to deal with compatibility etc, but I think it\u0027s worth it, and I think that a"},{"line_number":115,"context_line":"    rolling driver model might help alleviate some of the problems I see with"},{"line_number":116,"context_line":"    Cinder development."},{"line_number":117,"context_line":""},{"line_number":118,"context_line":"Data model impact"},{"line_number":119,"context_line":"-----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_325aa984","line":116,"updated":"2015-06-05 17:45:16.000000000","message":"Another advantage to having a 3rd party repo with all the drivers, is that we can still peer review code and learn from each other.   \n\nOne of the advantages that I\u0027ve seen for vendors that support both iSCSI and FC, is that those new drivers that come in often model their driver structure and implementations after existing drivers that do the same thing.   Does Cinder itself care about this?  No, but folks that are writing drivers do.   \n\n  It\u0027s a bit harder to do that when vendor plugins are spread around the internets and not peer reviewed.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":2243,"name":"John Griffith","email":"john.griffith8@gmail.com","username":"john-griffith"},"change_message_id":"b2c3d87fe03ac3d19ff2ba0b100cfc05684b3fc0","unresolved":false,"context_lines":[{"line_number":113,"context_line":"    of updating it and pushing changes out.  This will require some effort obviously"},{"line_number":114,"context_line":"    to deal with compatibility etc, but I think it\u0027s worth it, and I think that a"},{"line_number":115,"context_line":"    rolling driver model might help alleviate some of the problems I see with"},{"line_number":116,"context_line":"    Cinder development."},{"line_number":117,"context_line":""},{"line_number":118,"context_line":"Data model impact"},{"line_number":119,"context_line":"-----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_c3aa6919","line":116,"in_reply_to":"3afb71cf_ef5e6e44","updated":"2015-06-05 19:11:19.000000000","message":"This might be the best way to start, again it assumes you have enough qualified/willing Lieutenants, but I\u0027m game... fair compromise.\n\nThen we can revisit the topic again in six months if there\u0027s any interest.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":5997,"name":"Walt","display_name":"Hemna","email":"waboring@hemna.com","username":"walter-boring","status":"SAP"},"change_message_id":"f3f703b0ca845d34bbc88aef843dcdcddaad8110","unresolved":false,"context_lines":[{"line_number":148,"context_line":"Other deployer impact"},{"line_number":149,"context_line":"---------------------"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"Plugins would have to be installed via \"pip install some-plugin\""},{"line_number":152,"context_line":""},{"line_number":153,"context_line":"Developer impact"},{"line_number":154,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_12720de6","line":151,"updated":"2015-06-05 17:45:16.000000000","message":"One of the single biggest advantages of this is allowing vendor drivers to be upgraded for an OpenStack release that is no longer supported.\n\nCurrently today, in tree drivers are only updateable via the backport process.  Most of the time simple bug fixes for drivers never make it into previous releases, because they must fit certain \u0027critical core fixes\u0027 criteria.   \n\nWhen we allow drivers to be installed via the standard pypi, pip mechanism, vendor drivers can now be upgraded for an OpenStack release long after it\u0027s not supported.\n\nSo, say we register the driver with a range of values.\n3pardrivers\u003e\u003d1.0.0,\u003c\u003d2.0.0\n\nNow any 1.X.X fixes will get picked up by a deployment for that version of Cinder.\n\nI think this is a huge benefit to operators, vendors and end users that doesn\u0027t exist today easily.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":2243,"name":"John Griffith","email":"john.griffith8@gmail.com","username":"john-griffith"},"change_message_id":"b2c3d87fe03ac3d19ff2ba0b100cfc05684b3fc0","unresolved":false,"context_lines":[{"line_number":148,"context_line":"Other deployer impact"},{"line_number":149,"context_line":"---------------------"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"Plugins would have to be installed via \"pip install some-plugin\""},{"line_number":152,"context_line":""},{"line_number":153,"context_line":"Developer impact"},{"line_number":154,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_436d5903","line":151,"in_reply_to":"3afb71cf_12720de6","updated":"2015-06-05 19:11:19.000000000","message":"Yup, the compromise proposal is intended to provide this capability as well FWIW.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":5997,"name":"Walt","display_name":"Hemna","email":"waboring@hemna.com","username":"walter-boring","status":"SAP"},"change_message_id":"f3f703b0ca845d34bbc88aef843dcdcddaad8110","unresolved":false,"context_lines":[{"line_number":156,"context_line":"Significant impact on developers, contributors to Cinder would now actually"},{"line_number":157,"context_line":"be making contributions to Cinder itself, and those that just submit to their"},{"line_number":158,"context_line":"own special interest or driver can do it elsewhere without having to deal with"},{"line_number":159,"context_line":"process."},{"line_number":160,"context_line":""},{"line_number":161,"context_line":"Implementation"},{"line_number":162,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_323bc9ed","line":159,"updated":"2015-06-05 17:45:16.000000000","message":"Are we going to have a \u0027registry\u0027 of drivers in cinder, say a driver.yaml that vendor drivers can put ranges of their versions and point to their repo where the source exists?","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"},{"author":{"_account_id":2243,"name":"John Griffith","email":"john.griffith8@gmail.com","username":"john-griffith"},"change_message_id":"b2c3d87fe03ac3d19ff2ba0b100cfc05684b3fc0","unresolved":false,"context_lines":[{"line_number":156,"context_line":"Significant impact on developers, contributors to Cinder would now actually"},{"line_number":157,"context_line":"be making contributions to Cinder itself, and those that just submit to their"},{"line_number":158,"context_line":"own special interest or driver can do it elsewhere without having to deal with"},{"line_number":159,"context_line":"process."},{"line_number":160,"context_line":""},{"line_number":161,"context_line":"Implementation"},{"line_number":162,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3afb71cf_631df572","line":159,"in_reply_to":"3afb71cf_323bc9ed","updated":"2015-06-05 19:11:19.000000000","message":"Yeah, I would\u0027ve done something like that... but that\u0027s one reason the compromised suggestion might be better here.\n\nThe trouble is that gets all sorts of goofy with imports in the drivers of modules.  I thought I had a good way to get around that but frankly it may prove as I\u0027ve been kinda jaded in the past that \"adding repos\" is almost always a big fat PITA.","commit_id":"6e29d3041a7617c2c1b15c3458bed19e1fa458fa"}]}
