Blog from March, 2007

Currently the descriptions are absent and so we are setting the code to be the description.

This is our impl of getEnrollmentStatusDescriptions:

	@SuppressWarnings("unchecked")
	public Map<String, String> getEnrollmentStatusDescriptions(Locale locale) {
		//TODO: setup the NLS props on the connection based on locale?
		List<String> statuses = getHibernateTemplate().findByNamedQuery("findEnrollmentStatusDescriptions");
		
		//TODO: do we need human-readable descriptions here?
		Map<String, String> rv = new HashMap(statuses.size());
		for(Iterator i = statuses.iterator();i.hasNext();) {
			String code = (String)i.next();
			rv.put(code, code);
		}
		
		return rv;
         }

CourseSets play prominently in assumptions written into the CMS API (comments, at least)

CourseManagementService.isEmpty(String courseSetEid) for example assumes that the String param is a eid for a CourseSet and the OOTB hibernate implementation runs this HQL:

from CourseSetCmImpl as cs where cs.eid=:eid and
        	(size(cs.canonicalCourses) > 0 or size(cs.courseOfferings) > 0)

We have Meeting and Section implemented and mapped now.

So, before we can get a full CourseManagementService implementation that works, we'll need to decide what to do with some some remaining object mappings

There are other things to decide too.

In the meantime, I am overriding the parent class implementations of unimplementable methods and logging errors

Here are some of those methods. Do we need these? I don't know.

There's a method that uses the courseset eid along with the academic session eid to get courseofferings.

public Set findCourseOfferings(String courseSetEid,
			String academicSessionEid)

We will also need memberships in order to implement

public Map findCourseOfferingRoles(String userEid) and 
public Map findCourseSetRoles(String userEid)

...since roles are only in memberships

WRT CourseSets

public List<CourseSet> findCourseSets(final String category)

the String could be anything. This could be coded as well: "dept:chemistry", "loose affiliation: extension"

The key to me is to not think of CourseSets as having to statically point to specific named node in our enterprise hierarchy (like 'School') but can be qualified a bit.

WRT Grading Schemes

public Map<String, String> getGradingSchemeDescriptions(Locale locale)

I suppose this will be easy once we figure out how to get Membership objects impl'd

And then there's and sql thing....

public Set<EnrollmentSet> findCurrentlyEnrolledEnrollmentSets(final String userId) 

... currently uses this in the named query:

'select enr.enrollmentSet from EnrollmentCmImpl as enr where enr.userId=:userId and enr.dropped=false and
        	(enr.enrollmentSet.courseOffering.startDate is null or enr.enrollmentSet.courseOffering.startDate <= current_date()) and
        	(enr.enrollmentSet.courseOffering.endDate is null or enr.enrollmentSet.courseOffering.endDate >= current_date())'

... which would translate in one case to 'any user who has not dropped and belongs to an enrollmentset that has no start and no end date.'

I think we can safely remove the null checks here but ?

There's a similar issue with

 
public Set<EnrollmentSet> findCurrentlyInstructingEnrollmentSets(final String userId)

WRT sections:

public Set findEnrolledSections(String userEid)

public Set findEnrolledSections(String userEid, String academicSessionEid)

And, this needs sections and Memberships:

public Map findSectionRoles(String userEid)

..that's all for now

There is currently no way to get from an Enrollment to an EnrollmentSet in the CM API. The Sakai OOTB hibernate implementation has an EnrollmentSet object as a property to each Enrollment impl. As Josh has pointed out in emails before, this is a heavy memory object map. I proposed to Josh that he add a setter and getter for a String property for the EnrollmentSet's EID into the Enrollment interface. He said that he would look to add it for Sakai 2.5.

Thomas:
Hardware meeting discussion about using braddock as third node in prod cluster. This week will follow the old off-line maint. protocol.
414 will roll to prod.
applying mail re-routing, testing virus interception this PM (joncarlo, et al.)
2.4 release planning meeting:

James:
sak-524 assignment-gradebook thing. and sak-525.

Mike:
into the stats tool now.... making implementation classes now.

Lisa: out

Brian:
had a break thru on solving the CAS memory leak issue: file i/o blocked, and orphaned, threads.
working also on creating source in external db from whitepages for displayName in Sakai

Jon:
CM....mapped a few more things, created a test skel.

We pretty much just chatted for about 10 minutes......

Brian:
new mviews all psyched cuz they're complex types and we can get indexes on stuff

James:
Bug from Training and support meeting: assignments created before migration and linked to gradebook of old data are getting orphaned in some way
Needs mac and non admin user to test webdav

Mike:
More learning.... close to being into the Stat's tool work again.

Thomas:
Siteinfo asdd participants not notified issue: config file setting, got UserDirectoryProviders to work in trunk
meeting tomorrow 3-4 about consolidating subversion conference call in Central

Jon:
Course management mapping - -showed mapping and talked some about the current state and issues

there should be a limited list of term type available in the course site creation wizard and so should be joined with active term tpes for the user

James:
Fixing JIRA bugs 516, 334 (8226) and installing breeze.

Thomas:
asked about the configuration

Thomas:
Trying to compile Sakai.... test-harness and ucd-distauth not working, Josh Holtsman looking into test-harness
only user-based impl remains in ucd-dataservice
Brian:
Using netbeans plugin called profiler plugin for netbeans to profile the CAS stuff. Will do vendor drop tests with src code with Emily today and will also create test mview for Jon to hit with cm-ucd-impl.

Lisa:
Test plan submitted to Sakai Foundation QA.

Jon:
Filling out SCORM docs on fnd confluence waiting for an mview to work with (Brian)

Mike:
progressing, jsf and jsp learning.

The ideal situation would be to have an opaque id generated by the process that updates the tables (I am thinking mviews here) this opaque Id can be indexed, one would think, and hibernate would be cheery that it had a real id, and it would be zippy to search.

Question (TPA):
We are in the process of implementing CM and well as the cm-authz-provider. It almost seems like that the cm-authz-provider implementation could be general, usable by anybody, since it only makes calls to the CM API. Then if an institution implements/adjust the CM implementation, we wouldn't have to modify the cm-authz-provider and use it OOTB. Is that correct?

Answer (from Josh):
You are correct. There's no need to modify the group provider... just implement CM and configure the OOTB CM-based group provider in its components.xml file. More documentation will be coming, but the configuration is intended to be flexible and straight forward.

The older CM implementation objects used eid's in a way that did not ensure their uniqueness: CourseInfo object used eid=CRN and the CRN is not unique across all terms. The queries ensured the uniqueness at the DAO level by qualifying, in this case, with the TERM_CODE.

A unique eid would be in this case CRN+'_'+TERM_CODE', as one possibility.

But, we would lose the ability to take advantage of the underlying BANNER indexes in the tables on which our views are based if we tried to populate view columns with a concatenate as in select concat(CRN, TERM_CODE) as eid, ....

Composite Id's- with and with out component classes
Composite Id's using component classes seems to require that the classes can be written somewhere, since they must be serializable.
Not using component classes seems straighrfoward enough but

Not sure if a separatly mapped object can be used for this (in another table). That would work. But, since composite Id's seem to be generally frowned upon in hibernate, the next best solution would be to extend net.sf.hibernate.id.IdentifierGenerator and user that as a generator class.

Experimentation ensues.

Lisa:
filed SAK-512
– how to handle bugs discussion ensued
— we talked about PETA, cloning....GOSH Lisa!

[Tried cloning issues and liked it. Trouble tickets will be cloned and moved to Stories when appropriate and tickets sent back to ITX]

Mike: JSF hibernate and Spring... claims he's learning alot.

James: swimming and hiking.

Thomas: svn ucd-sakai trunk, practicing vendor drop stuff.

Mad TV

Jon:
Coursemanagement - created project in svn will start making SPring JDBC impol of the CMService

This is the set of stories going back to the iteration committment meeting:

And this is the set of stories that appear to be our current backlog:

Strategy for SCORM Tool and in 2.5 Sakai: James and sketched out this approach today

1. CHTypes and CHH's
2. aop method interception for adl packages
   • allows for separation of code
3. specifications for a plug-able sequencing service
   • this is just for sequencing not for 'sequencing engine'
   • sequencing engine should be behind a WService
4. webservices for SCORM API
5. move applet to server - use AJAX client
6. strategy for SCORM1.2 content
7. implement Global Objectives
8. hibernate map remainder of SCODataManager object
   • maybe use this class as a DTO ?