Background
Paul Spaude 11:38
Here's an updated list of problems, goals, and a brief summary of takeaways from the call this morning. Again thanks to everyone for their thoughts and time, and thanks especially to Evolveum for creating the version we have today! Feel free to edit/add comments +
https://docs.google.com/document/d/1xFGmoWClFIBUz0Xaa4aO0d60UTJ4jYXkkWARSXsvS-c/edit?usp=sharing
Chris Hyzer 15:49
i just had a good talk with JJ and my opinion as I have said before about the midpoint integration is Grouper should be pushing to midpoint.
- Provisioning Target is from a single-assign metadata in Grouper
- The provisioner could map whatever it wants to group name, display name, id_index, description
- id_index will generally be a numeric bigint that can be used for efficient foreign keys
- The group table has some common columns that can be used or not
- When Grouper changed a record it will edit the last_modified column which can be used as a change log
- Data will not be deleted at first, but rather will be set with a deleted flag
- The T and F are strings of size 1 not boolean database types
- Group attributes will generally be single valued but could be multi-valued
- The User Subject_Id must be something that MidPoint can use to lookup a user, hopefully will be the Grouper Subject_Id
- Grouper can delete deleted rows from the table after a week (configurable)
- Two types of work will not happen at the same millisecond
- Discuss with Midpoint folks the edge case of reading tables and not seeing data in time
Some edits to Chris Hyzer’s original proposal based on Software Integration Working Group Discussions
gr_mp_groups
...
gr_mp_group_attributes
...
gr_mp_memberships
...
gr_mp_subjects
...
gr_mp_subject_attributes
...
This page was moved here