Websites/Mozilla.org/Archive/Design Implementation/RSS

Note: This page refers to the relaunch of the site completed in August 2009 and is available for reference.

What we want to do

Bug 493099: Hook up news feeds on new www.mozilla.org home page.

We want a feed parser that

• pulls latest feed info from feeds.mozilla.com for front page
• pulls latest feed info from news.rdf (for now) for front page
• optionally: creates mash ups of multiple feeds on sub-pages (e.g. Our causes > Education)

How to do it

Mozilla.org has a lot of visitors, so we probably want to cache feed items for some time to a) be nice to the feed provider and b) get better performance.

We probably want a database-based cache over a file-based cache, because it gives faster results.

TTL of the cached items can be relatively high (higher on the sub-pages than on the front page), because we offer the feeds to open up an additional window into the community and not to be a direct, always-up-to-date information resource.

Implementation

Overall ideas:

• Use a cron-job to generate static html with the news items
• Pull in the items from a db-based cache, refresh the cache occasionally

There are several different options (subject to evaluation).

Write a parser from scratch

• + maybe better performance
• + would hopefully be exactly what we want
• - more development time
• - not as well tested as other solutions

Zend Framework

• New BSD license
• very modular and flexible
• PHP5 only (read: makes use of new OOP features, which is good)
• well maintained, active development
• DB abstraction layer, cache handling
• some known issues, API a bit counter-intuitive in places
• probably not as heavily tested as other solutions (because relatively new)
• generally long turnaround time (bug filed to fix released)

PEAR

• XML_Feed_Parser
  • LGPL license
  • No maintainer
• XML_RSS
  • PHP license
  • No active development (last release 06)

SimplePie

• BSD license
• CURL extension required (need to check with reed, if it is enabled)
• Caching out of the box (flat file only right now, but could be overloaded with own DB-class)
• Meshing of multiple feeds out of the box
• Problems with some feeds, but workarounds exist
• Nice API
• Goodies such as feed auto-discovery (feels a lot like Firefox ;)
• Performance seems to be good with caching, but no idea, if it scales up nicely.

Magpie RSS

• GPL license
• No active development (last release 05)

PHP Universal Feed Parser

• GPL license
• seems to be a one-man effort
• uses OOP methodology