Grouper Users - Open Discussion List

[Liam Hoekenga <>]

I'm still trying to smash my brain against this grouper complex ("packed") attribute thing.

I think we'd prefer to use our IDM system as the authoritative source of the data in question.

We could use one of the gsh implementations and create an import script that processed our "people" entries as we wanted.. but that would be external to the actual loader mechanism.  I'd rather not have loaders running out of grouper's cron and system cron, and I'm not sure how it would even keep the correct state?  DROP / ADD?  Track the delta somehow?  (How do the actual grouper loaders keep track of this?)

wgthom suggested that maybe we pre-digest the data in LDAP and store it in easier to parse tables in SQL someplace.  My team has mixed feelings about that... and I ask, why would we do that instead of going back to the original source of that data (peoplesoft)?  Like I mentioned, tho, there's a strong preference to use our IDM as the source of authority.

For many of the fields in the packed data, there are LDAP containers that have the information used to decorate the packed data.  So, we've got containers with all of our department numbers, and job codes, and academic careers, etc.  Do we use the objects in those containers to create the reference folders, and then populate them by running multiple queries against the same attributes?

If we are asking for new loader feature(s), what is would actually be broadly useful?  A loader type that has the ability to pre-process the data it's handed.

More complicated JEXL?  (Could JEXL even do this?  I'm not familiar enough to say)

The ability for the loader to explode attribute values that are serialized objects into separate attributes and use those for LDAP_GROUPS_FROM_ATTRIBUTES?

Feed the attribute values to an ECMA script and let it do things?

I'm open to ideas.

thanks

Liam

On Fri, Apr 7, 2017 at 2:39 PM, Liam Hoekenga <> wrote:

On Tue, Apr 4, 2017 at 2:33 PM, Hyzer, Chris <> wrote:

We would need to add that

Which?  

"virtual attributes" in the loader?

un-stringify-ing an object into separate, usable attributes?