After mucking around with XDebug for a while (APD's pprofp seems to not like very large log files, as it just quits silently after thinking for a while), and discovering that XDebug 2-beta isn't exactly what I want (have to install KCacheGrind and stuff to view the logs, and I don't want to install any KDE libs), I decided that my stopwatch(); function would have to do for the time being.

Anyway, so indexing 5734 wiki entries varies between 140 - 180 seconds, I got 157.405181 seconds on my last run, and takes up close to 3MB of space on disk. Querying this index takes about half a second, as you can see in the attached file.

The Zend Framework contains a port of the Apache Lucene search project, written in PHP 5 http://framework.zend.com/manual/en/zend.search.html). I spent an hour or two over the weekend fiddling around with ZF and hacked together a very basic search engine, indexing data I had stored in a simple MySQL table. Overall, the ZF search class seems reasonable, the abstraction is nice, and it's really simple to get a index/query system up and running.

I didn't manage to change the default search fields, so searching for 'Michael' won't return any articles written by me, unless my name appears in the actual article text. To search for articles written by me, you have to specify 'author:Michael'. Probably a small issue, I must have a better look through the ZF source-code.

My documentation/planning page is available on the Moodle wiki at:
http://docs.moodle.org/en/Student_projects/Global_search

The SoC student project summary page is at:
http://docs.moodle.org/en/Student_projects

That's all for now, I have more reading to do.