> On March 30, 2018, 9:47 a.m., David Radley wrote: > > addons/models/0000-Area0/0010-base_model.json > > Line 205 (original), 205 (patched) > > <https://reviews.apache.org/r/66374/diff/1/?file=1990468#file1990468line205> > > > > I suggest there may be migration implications to renaming these base > > types. I suggest versioning this change.
Relationships 'dataset_process_inputs' and 'process_dataset_outputs' were introduced only in master branch in 1.0. Renaming ends of these relationships won't impact migration from earlier version. > On March 30, 2018, 9:47 a.m., David Radley wrote: > > addons/models/0300-Area3-SubjectArea/0310-Glossary.json > > Lines 89 (patched) > > <https://reviews.apache.org/r/66374/diff/1/?file=1990470#file1990470line89> > > > > I thought that we were thinking of these constraints as legacy and had > > assumed that we would not be adding any new ones. Doesn't the owned > > constraint imply that this is a one directional relationship, but the > > relationship definition is bidirectional. > > > > What does it mean that there is an attribute that uses the contraint > > but is not marked as legacy? > > > > I am wondering what we gain by adding the attribute in this way? - Without mandatory attribute 'anchor', terms and categories can exist without being associated with a glossary. This is incorrect. - Current relationship model doesn't allow to mark a relationship attribute as manadatory. In fact allowing such constraint on relationships will make it difficult/impossible to build clients that deal with CRUD of entities in such relationships, as they need to be updated whenever a new mandatory relationship attribute is introduced later. - From modelling point of view, attributes like 'anchor' and 'parent' are integral part of Term and Category; and these shouldn't have to be injected via relationship > On March 30, 2018, 9:47 a.m., David Radley wrote: > > addons/models/0300-Area3-SubjectArea/0320-CategoryHierarchy.json > > Lines 32 (patched) > > <https://reviews.apache.org/r/66374/diff/1/?file=1990471#file1990471line32> > > > > I wonder why we are adding legacy attributes. I can see it gives us a > > manditory parent; I thought the relationship design was such that we were > > dropping this capability. The field name 'legacy' is unfortunate (I think you pointed this earlier when this was introduced). It is meant to convey if the entity at this relationship end has an attribute that represents this end. - Madhan ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/66374/#review200231 ----------------------------------------------------------- On March 30, 2018, 8:16 a.m., Madhan Neethiraj wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/66374/ > ----------------------------------------------------------- > > (Updated March 30, 2018, 8:16 a.m.) > > > Review request for atlas. > > > Bugs: ATLAS-2526 > https://issues.apache.org/jira/browse/ATLAS-2526 > > > Repository: atlas > > > Description > ------- > > - fixed incorrect syntax/braces in 0120-Collections.json that resulted in > relationshipDefs not to be loaded > - renamed sourceToProcesses as inputToProcess and sinkFromProcesses as > outputFromProcess > - added terms and categories as direct attributes of Glossary > - added anchor, parentCategory and childrenCategories as direct attributes of > GlossaryCategory > - added anchor as direct attribute of GlossaryTerm > > > Diffs > ----- > > addons/models/0000-Area0/0010-base_model.json aebe955d > addons/models/0100-Area1-Collaboration/0120-Collections.json 4dbd2adb > addons/models/0300-Area3-SubjectArea/0310-Glossary.json cac2ce2a > addons/models/0300-Area3-SubjectArea/0320-CategoryHierarchy.json 10637970 > addons/models/0300-Area3-SubjectArea/0330-Terms.json f492ddfd > > > Diff: https://reviews.apache.org/r/66374/diff/1/ > > > Testing > ------- > > Verified Atlas server starts up successfully with updated models > > > Thanks, > > Madhan Neethiraj > >
