20070423 Monday April 23, 2007

How does the FSM work?

There's the poetic answer. There's also an answer that is a bit more technical.
  1. Every 24 hours (and once an hour on Thursdays), the FSM Loader runs. It looks for new Farktography contests listed on the main page, retrieves the contest page and the result page and extracts the entries that have received votes. This phase makes heavy use of HTML Parsers, XPath expressions, and various networking and file handling routines.
  2. The raw entries and the contest are loaded into the FSM Database, which used to be served by Derby, but is now MySQL 4.1.?.
  3. To pickup late votes (votes that come in after 24 hours), the past 4 contests are re-retrieved from Fark.com and reprocessed. Any vote that comes in after 4 weeks won't be counted until I feel like rebuilding the entire DB, if at all. Fark.com archives the threads and turns off voting sometime before the 4 week window closes
  4. After all votes have been accounted for, summary tables are built to assist with the reporting
  5. The static reports are then generated by BIRT, an open-source reporting tool.
  6. The reports and the FSM main page are then assembled (using Velocity) and deployed onto Schneeworld.com
Posted by schnee ( Apr 23 2007, 10:58:43 AM CDT ) Permalink
Linkbacks:
Comments:

Post a Comment:

Comments are closed for this entry.