Skip to Content.
Sympa Menu

grouper-dev - Draft Minutes: Grouper Call 5-March-2008

Subject: Grouper Developers Forum

List archive

Draft Minutes: Grouper Call 5-March-2008


Chronological Thread 
  • From: Steve Olshansky <>
  • To: Grouper Dev <>
  • Subject: Draft Minutes: Grouper Call 5-March-2008
  • Date: Tue, 11 Mar 2008 15:08:57 -0600

Draft Minutes
*Grouper Call: 5-March-2008*

*Attending*
Tom Barton, U. Chicago (chair)
Kathryn Huxtable, independent
Sanjay Vivek, Newcastle U. (UK)
Gary Brown, Bristol U. (UK)
Shilen Patel, Duke
Joy Veronneau, Cornell U.
Bert Bee-Lindgren, Georgia Tech U.
Steve Barrett, Cornell U.
Chris Hyzer, U. Penn
Mike Olive, Stanford
RL "Bob" Morgan, U. Washington
Steve Olshansky, Internet2 (scribe)

**New Action Items**
[AI] (Tom) will consult with Blair about duplicate columns of IDs/UUIDs
[AI] (SteveB and Chris) will sidebar about Simple and non-Simple SOAP Web Services, and report back to the group.


*Carryover Action Items*
[AI] (TomZ) will start a discussion about the potential applicability of the Shibboleth extension framework on the Grouper-Dev list
[AI] (UI Subgroup/Jennifer) will offer some corrections to the UI, and some additional pages
[AI] (Chris and Shilen) will investigate performance degradation issues and report back to the group.
[AI] (All) review open JIRA items and discuss via the list as inclined.

**Discussion**

*Updates*
UI
April is looking likely for a next release, thus we ought to be aiming for a feature freeze in the near future, and some update to the UI is likely to be included. The subgroup reported that they are working on a recommended set of terms for Grouper concepts, and are working on incorporating them into the relevant screens for wider review.

API
Chris and Shilen investigated performance degradation issues and reported that so far there is no apparent reason why the use of Hibernate3 would lead to ~10% slower performance, although the performance problems do appear to be related to this. Work is continuing to isolate the problem, as there were some features added that also contributed to overhead and thus there is a need to roll back to the point where Hibernate3 was the only new factor. They are still investigating and will report back.

Implementing foreign keys and corresponding Hibernate mappings may be desirable for the next release. A key objective is to make Grouper independent of RDBMSs. There was some discussion about whether a change of this magnitude is a good use of our time, in light of other priorities. Apache DdlUtils was also suggested as worth investigating in this context.
http://db.apache.org/ddlutils/

There is an underlying need to ensure that group identifiers will be persistent, even if the group names change. UUID may be the best option for this. Is there also a need for IDs as well?

[AI] (Tom) will consult with Blair about duplicate columns of IDs/UUIDs

Searching is also an area that could do with some performance improvement. Performing complex queries would be more efficient, v. doing multiple single queries and presenting the union as the result.

Web Services/SOAP
Chris is looking at how best to approach this in our context. Once he changes clients to support this, he will work on "groups-lite" on top.

Given that there is Simple SOAP, non-Simple SOAP, and Lite (not SOAP) what distinct application scenarios are most appropriate for each? Would it make sense for us to start simpler and get more complex in response to demand, rather that implementing complexity in the expectation of need? Once the functionality is in there it will need to be supported. Is there really value in having a simple method for each web service call?

[AI] (SteveB and Chris) will sidebar about Simple and non-Simple SOAP Web Services, and report back to the group.

Groups-Lite (formerly RESTful Web Service)
See mail from RL "Bob" to the dev list 5-March-2008 regarding updates in the wiki.
https://wiki.internet2.edu/confluence/display/GrouperWG/Groups-Lite+Web+Services

Ldappc
Kathryn is performing performance tests on traditional Ldappc, the Brown approach, and using an intermediate DB table. The Brown approach looks promising, and Kathryn is modifying that to improve performance even more. Her work continues and she will report back to the list soon. Her early estimate is that 4 weeks would be required to complete this work.

Next call will be Wed 19-March Noon EDT.


--
Steve Olshansky
Internet2 Middleware & Security Flywheel
http://middleware.internet2.edu/
http://security.internet2.edu/


  • Draft Minutes: Grouper Call 5-March-2008, Steve Olshansky, 03/11/2008

Archive powered by MHonArc 2.6.16.

Top of Page