Hot damn, it looks like I might actually finish the creation of Priblog that I’ve always been talking about. While I really like the idea of working with flat XML files, the fact of the matter is that performance matters to me. I hate how slow pyBlosxom becomes when it has to walk all of my entries. The additional overhead of firing off a python process on every page hit isn’t that great either. I’m sure I could find a way to get around that using mod_python or something like that, but I think that will just raise more problems.
Instead, I’ve been playing with (cue Warrant music here) CherryPy. Basically, CherryPy is a persistant python process that serves up web pages. This means that I can maintain one set of database connections, one listing of the entries, and all that jazz. It’s terribly easy to create nice web applications with it. Add in built-in session management, and you can see why I’m happy.
Basically the new version of Priblog works with a PostgreSQL backend. It requires LibXML to generate the XML pages and do some fun stuff with XSLT for browsers that can’t handle it, like IE. Right now, I’d estimate that I’m at about 50% of the progress that I have in my pyBlosxom install. I still need to add in category browsing and date browsing, but those are the last outstanding features. It looks like this time I might actually finish something. Yay.
I half feel like I’m now in a race with another guy. Apparently there are others who wish to accomplish the same. I guess mine won’t truly be a pyBlosxom, because for time being you’re going to need a database behind it. Not sure if I’ll keep that idea forever or not.