FreeRoller update.
A number of Roller performance/load fixes have been applied to FreeRoller over the last month or so including: persistence session per request, upgrade to OSCache 2.0-beta2, enabling OSCache unlimited disk cache, and upping the page cache timeouts from 1 to 3 hours. The unlimited disk cache was enabled this morning and the performance and load improvements seem to be pretty significant, but there are still reports of some 404 errors that need to be investigated.
If you display referers on your site, you'll notice the higher page cache timeouts because your referers list will only be refreshed every three hours (unless you post or manually flush your cache).
Links.
- Via Mike Cannon Brookes: Erik Hatcher says good-bye to the Struts-Dev list and moves on to Tapestry for his next project.
- Via Simon Bruning: side-by-side comparison of Java and Python
- StrutsTestCase Tutorial at Java Boutique: Shows you how to unit test your Struts actions using mock objects via StrutsTestCase.
- Dan Gillmore: Microsoft is hiring developers to work on blogging software for MSN and for "not product specific' projects.
- Robert X. Cringley: off-shoring IT jobs is tantamount to age-discrimination and suicidal for US companies.
- Rafe hears that natively compiled Eclipse is "blazingly fast": I need to try this on my Redhat homebox.
- Shifted Librarian: Roller Stone allows you to create RSS feeds for your favorite artists.
Clearly, something is wrong with Daypop.
I knew something was amiss when yesterday my four year old WFC vs. JFC article rose up into the Daypop 15 Top News Stories. My day-popularity is still going strong today. At this moment, 4 of the 15 Top Posts on Daypop point to my weblog entries. Clearly, something is wrong with Daypop.
What's the problem? The other day I tweaked the layout of my site the other day and,apparently, the changes that I made sent irritating ripples through Daypop. Daypop tracks the most popular news stories and weblog posts by counting the number of weblogs that link to those news stories and weblog posts. By changing my site layout, I effectively changed the contents of hundreds of web pages on my site. Daypop's flawed top post algorithm saw this as a giant burst of day-popularity for the various links that decorate the right hand side of my weblog.
Why doesn't Daypop account for this type of site-layout change, I mean, don't layout changes like this happen all the time in the world of weblogs? I wonder: is there something that I could have done to keep my changes from driving Daypop bonkers?