)]}'
{"specs/backlog/validated-and-indexed-extra-fields.rst":[{"author":{"_account_id":5538,"name":"Rushi Agrawal","email":"rushi.openstack@gmail.com","username":"rushiagr"},"change_message_id":"893ba87e76141bc6d496a0d543f0ef562d241e1b","unresolved":false,"context_lines":[{"line_number":21,"context_line":"Problem Description"},{"line_number":22,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":23,"context_line":""},{"line_number":24,"context_line":"Deployers and Operators utilize the \"extra\" fields for a number of use cases"},{"line_number":25,"context_line":"such as billing information, data about what business unit a resource belongs"},{"line_number":26,"context_line":"to and much more. There have been many cases that these operators and deployers"},{"line_number":27,"context_line":"want to be able to validate these fields and/or searching based upon indexing"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_52be2b00","line":24,"updated":"2015-06-22 18:24:26.000000000","message":"How do they do it? Do they access the db directly, or does keystone provide a way for people to put/manipulate \u0027extra\u0027 data in the db via the api?","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"780db64e3c06522d24c33244456e0174c8a15d2d","unresolved":false,"context_lines":[{"line_number":21,"context_line":"Problem Description"},{"line_number":22,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":23,"context_line":""},{"line_number":24,"context_line":"Deployers and Operators utilize the \"extra\" fields for a number of use cases"},{"line_number":25,"context_line":"such as billing information, data about what business unit a resource belongs"},{"line_number":26,"context_line":"to and much more. There have been many cases that these operators and deployers"},{"line_number":27,"context_line":"want to be able to validate these fields and/or searching based upon indexing"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_522279a4","line":24,"in_reply_to":"fa32b979_52be2b00","updated":"2015-06-23 15:39:15.000000000","message":"The way it works now, if anyone passes random attributes to keystone that are not part of the model, they are stored as \"extra\" in a serialized object.\n\nThe change here is that we would provide JSON Schema validation for the attributes deployers actually care about and reject random values not defined when this system is uesed.","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":5538,"name":"Rushi Agrawal","email":"rushi.openstack@gmail.com","username":"rushiagr"},"change_message_id":"893ba87e76141bc6d496a0d543f0ef562d241e1b","unresolved":false,"context_lines":[{"line_number":47,"context_line":"   This will have no impact on the first-class attributes defined across"},{"line_number":48,"context_line":"   all Keystone deployments (such as `username` for users)."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"2. At the definition time of the extra attribute, the deployer/operator could"},{"line_number":51,"context_line":"   specify the attribute as indexed. This would signal to Keystone that it"},{"line_number":52,"context_line":"   would need to internally manage the index (in a separate table or similar)"},{"line_number":53,"context_line":"   of the attribute for the resource type."}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_253a3f99","line":50,"updated":"2015-06-22 18:24:26.000000000","message":"Does this also mean the \u0027create users\u0027, \u0027create domains\u0027 etc api will be impacted? If so, would you mind providing more details as to how the rest api would look like? I mean, we\u0027ll need some way to specify that these few parameters are \u0027extra\u0027, and their type too, via a json..","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"780db64e3c06522d24c33244456e0174c8a15d2d","unresolved":false,"context_lines":[{"line_number":47,"context_line":"   This will have no impact on the first-class attributes defined across"},{"line_number":48,"context_line":"   all Keystone deployments (such as `username` for users)."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"2. At the definition time of the extra attribute, the deployer/operator could"},{"line_number":51,"context_line":"   specify the attribute as indexed. This would signal to Keystone that it"},{"line_number":52,"context_line":"   would need to internally manage the index (in a separate table or similar)"},{"line_number":53,"context_line":"   of the attribute for the resource type."}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_d2452943","line":50,"in_reply_to":"fa32b979_253a3f99","updated":"2015-06-23 15:39:15.000000000","message":"All the apis that support \"extra\" values are affected. This would be a way that deployers can limit what gets shoved into extra (the whole extra design is poor, but we can\u0027t get rid of it because it is used by a large number of deployers).","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":5538,"name":"Rushi Agrawal","email":"rushi.openstack@gmail.com","username":"rushiagr"},"change_message_id":"893ba87e76141bc6d496a0d543f0ef562d241e1b","unresolved":false,"context_lines":[{"line_number":49,"context_line":""},{"line_number":50,"context_line":"2. At the definition time of the extra attribute, the deployer/operator could"},{"line_number":51,"context_line":"   specify the attribute as indexed. This would signal to Keystone that it"},{"line_number":52,"context_line":"   would need to internally manage the index (in a separate table or similar)"},{"line_number":53,"context_line":"   of the attribute for the resource type."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"   A new query param will be added (based upon this definition) for the defined"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_72f7af59","line":52,"updated":"2015-06-22 18:24:26.000000000","message":"I think there is room for a bit more detail regarding how the tables will be laid out in the db. If it\u0027s the same table, are we going to create a few, say 2-3 \u0027reserved\u0027 columns for extra attributes, so as to assist in indexing? Similarly what about when we have separate table? How are we going to manage with the fact that some deployers might provide no attribute (in that case, those columns will be wasted, which might be okay I guess), or some might want to index on say half a dozen attributes (and clearly we can\u0027t have those many columns to index on)?","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"780db64e3c06522d24c33244456e0174c8a15d2d","unresolved":false,"context_lines":[{"line_number":49,"context_line":""},{"line_number":50,"context_line":"2. At the definition time of the extra attribute, the deployer/operator could"},{"line_number":51,"context_line":"   specify the attribute as indexed. This would signal to Keystone that it"},{"line_number":52,"context_line":"   would need to internally manage the index (in a separate table or similar)"},{"line_number":53,"context_line":"   of the attribute for the resource type."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"   A new query param will be added (based upon this definition) for the defined"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_729c5dc0","line":52,"in_reply_to":"fa32b979_72f7af59","updated":"2015-06-23 15:39:15.000000000","message":"This is again, already addressed, this is a question of how the index (look up reverse indexes) are built. This is not reserved columns.","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":7191,"name":"Jamie Lennox","email":"jamielennox@gmail.com","username":"jamielennox"},"change_message_id":"a8112f7e157aa840b6abd41ef35c0a9d83f48385","unresolved":false,"context_lines":[{"line_number":56,"context_line":"   resource-type (e.g."},{"line_number":57,"context_line":"   `/users?\u0026extra_attr\u003d{attr_name}\u0026value\u003d{attr_value}`). This will"},{"line_number":58,"context_line":"   allow the user(s) of Keystone to search for a resource that has the defined"},{"line_number":59,"context_line":"   attribute and (optionally) attribute value."},{"line_number":60,"context_line":""},{"line_number":61,"context_line":"Alternatives"},{"line_number":62,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_db6f29c2","line":59,"updated":"2015-06-11 08:29:24.000000000","message":"I don\u0027t think we should support indexes - at least initially. This is adding new functionality to the feature because we certainly don\u0027t support it now. \n\nI want this to be us informing people they should use it less not that we want to support there case because then we\u0027ll never get rid of it.","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"2557b0a4ddc6dacbd7e17866073a4415c1a2efa9","unresolved":false,"context_lines":[{"line_number":56,"context_line":"   resource-type (e.g."},{"line_number":57,"context_line":"   `/users?\u0026extra_attr\u003d{attr_name}\u0026value\u003d{attr_value}`). This will"},{"line_number":58,"context_line":"   allow the user(s) of Keystone to search for a resource that has the defined"},{"line_number":59,"context_line":"   attribute and (optionally) attribute value."},{"line_number":60,"context_line":""},{"line_number":61,"context_line":"Alternatives"},{"line_number":62,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_9999b6c6","line":59,"updated":"2015-06-11 10:30:32.000000000","message":"The ask, more than validation, has been for the search/indexing of the extra attrs. We can\u0027t really get rid of it, we should make using it better. Searching through every single resource in keystone (e.g. every project) to know what a specific attr is (or if it is even seT), is really really really bad. I\u0027d like to get the index to be tied to requiring validation - so we can push away from it being a random bundle of whatever people shove into the resource.","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":5538,"name":"Rushi Agrawal","email":"rushi.openstack@gmail.com","username":"rushiagr"},"change_message_id":"893ba87e76141bc6d496a0d543f0ef562d241e1b","unresolved":false,"context_lines":[{"line_number":61,"context_line":"Alternatives"},{"line_number":62,"context_line":"------------"},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"An  alternative to having a CRUD interface for defining the attributes is to"},{"line_number":65,"context_line":"make the important attributes a core part of the resource objects themselves."},{"line_number":66,"context_line":"The issue with making these extra attributes a first-class part of the resource"},{"line_number":67,"context_line":"is that it requires either direct customization by the deployer (changing"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_85954b41","line":64,"updated":"2015-06-22 18:24:26.000000000","message":"\"Adding a separate api for extra-attribute manipulation, which is kept separate from the core resource api\" --\u003e another alternative I can think of. I see some value in going this way.. This looks like a clean way of keeping this indexing-and-querying-extra-attributes work from the core APIs","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"780db64e3c06522d24c33244456e0174c8a15d2d","unresolved":false,"context_lines":[{"line_number":61,"context_line":"Alternatives"},{"line_number":62,"context_line":"------------"},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"An  alternative to having a CRUD interface for defining the attributes is to"},{"line_number":65,"context_line":"make the important attributes a core part of the resource objects themselves."},{"line_number":66,"context_line":"The issue with making these extra attributes a first-class part of the resource"},{"line_number":67,"context_line":"is that it requires either direct customization by the deployer (changing"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_72d1bd56","line":64,"in_reply_to":"fa32b979_85954b41","updated":"2015-06-23 15:39:15.000000000","message":"This is a drastic departure and suffers from all your previous comments of making the use-case complex.\n\nWe should *not* be adding top-level attributes or dynamic schema changes to the core tables for everyone\u0027s edge use cases.","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":7191,"name":"Jamie Lennox","email":"jamielennox@gmail.com","username":"jamielennox"},"change_message_id":"a8112f7e157aa840b6abd41ef35c0a9d83f48385","unresolved":false,"context_lines":[{"line_number":154,"context_line":"Documentation Impact"},{"line_number":155,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"Documentation of the new API(s) and query parameters will nbeed to be created."},{"line_number":158,"context_line":""},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"References"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3afb71cf_5bb3394d","line":157,"updated":"2015-06-11 08:29:24.000000000","message":"nbeed?","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"780db64e3c06522d24c33244456e0174c8a15d2d","unresolved":false,"context_lines":[{"line_number":154,"context_line":"Documentation Impact"},{"line_number":155,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"Documentation of the new API(s) and query parameters will nbeed to be created."},{"line_number":158,"context_line":""},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"References"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fa32b979_92d6c14d","line":157,"in_reply_to":"3afb71cf_5bb3394d","updated":"2015-06-23 15:39:15.000000000","message":"it\u0027s a typo :P","commit_id":"8e6a90c1dff94caa6f8375ae97c0d2de60ba1f86"},{"author":{"_account_id":7191,"name":"Jamie Lennox","email":"jamielennox@gmail.com","username":"jamielennox"},"change_message_id":"3f2b1e165e772eb67c85d7ff63a16b53fdcf8b98","unresolved":false,"context_lines":[{"line_number":54,"context_line":""},{"line_number":55,"context_line":"   A new query param will be added (based upon this definition) for the defined"},{"line_number":56,"context_line":"   resource-type (e.g."},{"line_number":57,"context_line":"   `/users?\u0026extra_attr\u003d{attr_name}\u0026value\u003d{attr_value}`). This will"},{"line_number":58,"context_line":"   allow the user(s) of Keystone to search for a resource that has the defined"},{"line_number":59,"context_line":"   attribute and (optionally) attribute value."},{"line_number":60,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"3a50d1a3_53136365","line":57,"updated":"2015-08-03 00:57:28.000000000","message":"This won\u0027t allow you to query on multiple values. \n\nSuggest like x_name\u003dvalue and anything with the x_ prefix is an addition.\n\nx_street\u003d\"hope\"\u0026x_city\u003d\"brisbane\"","commit_id":"fb2cf7b7b50d39ed1c01fd9e5f9e46ca09bcb296"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"c55a11a761f766bd0e8f92aa1de78b95d2be25b4","unresolved":false,"context_lines":[{"line_number":54,"context_line":""},{"line_number":55,"context_line":"   A new query param will be added (based upon this definition) for the defined"},{"line_number":56,"context_line":"   resource-type (e.g."},{"line_number":57,"context_line":"   `/users?\u0026extra_attr\u003d{attr_name}\u0026value\u003d{attr_value}`). This will"},{"line_number":58,"context_line":"   allow the user(s) of Keystone to search for a resource that has the defined"},{"line_number":59,"context_line":"   attribute and (optionally) attribute value."},{"line_number":60,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"da20952f_e7e0d3fc","line":57,"in_reply_to":"3a50d1a3_53136365","updated":"2015-08-28 04:15:15.000000000","message":"Yes, in the next revision.","commit_id":"fb2cf7b7b50d39ed1c01fd9e5f9e46ca09bcb296"},{"author":{"_account_id":1941,"name":"Lin Hua Cheng","email":"os.lcheng@gmail.com","username":"lin-hua-cheng"},"change_message_id":"452bcca64c79ffe5e1deacda2ffaf100c1f983c6","unresolved":false,"context_lines":[{"line_number":136,"context_line":"* Connect the CRUD defined JSON Schema Validators to the Controller(s) so that"},{"line_number":137,"context_line":"  validation can occur on the new attributes."},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"* Define/implement the mechanism for indexing the extra attributes"},{"line_number":140,"context_line":""},{"line_number":141,"context_line":"* Define/implement the API/Query parameters for querying based upon the extra"},{"line_number":142,"context_line":"  attributes."}],"source_content_type":"text/x-rst","patch_set":2,"id":"3a50d1a3_a691c79c","line":139,"updated":"2015-07-30 04:04:13.000000000","message":"What if the backend for the resource is LDAP, I assume indexing is not going to be supported?","commit_id":"fb2cf7b7b50d39ed1c01fd9e5f9e46ca09bcb296"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"928266b46bbeff352a0f0e5616f1b6a94c060ac4","unresolved":false,"context_lines":[{"line_number":136,"context_line":"* Connect the CRUD defined JSON Schema Validators to the Controller(s) so that"},{"line_number":137,"context_line":"  validation can occur on the new attributes."},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"* Define/implement the mechanism for indexing the extra attributes"},{"line_number":140,"context_line":""},{"line_number":141,"context_line":"* Define/implement the API/Query parameters for querying based upon the extra"},{"line_number":142,"context_line":"  attributes."}],"source_content_type":"text/x-rst","patch_set":2,"id":"3a50d1a3_14bbcc76","line":139,"in_reply_to":"3a50d1a3_a691c79c","updated":"2015-07-31 05:48:08.000000000","message":"Searching (the INDEX) is a SQL-ism, but LDAP search queries could support the same level of functionality without any crazy SQL-isms (searching a json-blob is highly inefficient, but the same search/query in LDAP can be quite efficient by comparison).","commit_id":"fb2cf7b7b50d39ed1c01fd9e5f9e46ca09bcb296"},{"author":{"_account_id":13997,"name":"SHIGEMATSU Mitsuhiro","email":"shigematsu.mitsuhiro@lab.ntt.co.jp","username":"pshige"},"change_message_id":"3219b8c3ce0503cb70d21ec42395a094e070b7a5","unresolved":false,"context_lines":[{"line_number":154,"context_line":"Documentation Impact"},{"line_number":155,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"Documentation of the new API(s) and query parameters will nbeed to be created."},{"line_number":158,"context_line":""},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"References"}],"source_content_type":"text/x-rst","patch_set":2,"id":"da20952f_0be4adde","line":157,"updated":"2015-08-30 03:27:23.000000000","message":"s/nbeed/need/","commit_id":"fb2cf7b7b50d39ed1c01fd9e5f9e46ca09bcb296"}]}
