Wednesday, April 20, 2011

The XMA Grows Up

At TEC this week we learned about a new MA framework coming to a FIM server near you!

The nickname for the new framework is ‘EZMA’.  My first take at this nickname was irony; since the framework provides was more functionality than the XMA SDK.  This raises the bar for the development expertise required to develop an MA using the new framework (it is REALLY easier?).  Raising this bar is a good thing IMHO since it makes it less enticing for non-developers to produce MAs, which is probably a good thing.

After going through the new SDK I realized it really is an EZMA because the added functionality means I no longer need to do a lot of things/hacks in my MA code.  For example, the call-based import functionality relieves me of the need to transform m objects into a file.  I have a lot of DSML code that now gets to retire!  Another example, on import I can now provide an update (partial object) instead of having to supply the complete attribute set.  This means that a lot of code in the OpenLDAP XMA can also retire since it has to chase changeLog entries into the actual directory entry to get the full attribute set.

Brian Desmond did an awesome job explaining this in his session today.  He’s done the early adopter work of building some MAs using the new framework and was able to tell us about it at the conference today.

No comments: