So our first CM load into Sakai worked just as expected. This is pretty slick and works so far just as expected.
My original xmlconvert.pl app was not very effecent and I was a bit anticipatory
during our roll over to the Fall 2008 data set.
It worked just fine, but the time to completion increased on basically a rocket like trajectory. The full 3k+ course convert to XML took over an hour (~80 minutes). Yikes! Sakai took roughly the same ammount of time to consume the data .. so not much change there.
The FIX!
While talking this over with a co-worker we formulated a better
way to do this than many,many,many file IO ops..
So _now_ I'm building the ~20mb xml file which contains all the course/enrollment data for IUP in around
17 seconds :-P
I basically build two hash-of-arrays which hold the data. The dept_course_sec is the key in both
of these HoAs while the array part contains either the instructor(s) of record or the enrolled students. Using
this data structure is so much more efficient it is not even funny. Why didn't I think of that in the first
place.. I guess I have alwase been a bit weary of using alot of hash types in perl .. but now... Im a very big
hash advocate.
Posted by Kevin Foote in sakai on August 28, 2008
