February 10th, 2004:

Our webservers are now running PHP5beta3 inside of Apache. As best as I can tell, this doesn’t seem to be causing any problems. In fact, it actually seems faster. However, I can’t test everything. If you see an error message of any type, on this site, or on your own sites, please send me a text message right away so I can put it back until I figure out what the problem is.

And I thought my comment spam was difficult to manage. Just look at the comments on this post.

PHP5′s Propel is an amazing library. It is based on Apache Turbine for Java. Well, it’s great in theory, anyway. I can’t run it, because I don’t run PHP5 currently. I’d love to, really. But, running PHP5 apps in production might not be the best idea when PHP5 is still beta. Additionally, if I were to write something like Inklog in PHP5, I would decrease my userbase, at least for now, drastically.

Regardless of those aspects, it looks great. The basics premise is this:

You describe your tables and their relationship to one another using XML files. A Phing build then converts that XML data into configuration files for use with Propel. as well as a bunch of PHP5 classes (yes, it actually writes the code) that are designed to access those tables using Propel. In the end, you end up with database tables that are directly related to PHP5 classes that allow objects to be stores and inflated at will. Additionally, you have an entire set of classes known as Peer classes, which allow you to retrieve the objects via a specific criteria.

Very powerful stuff. A large portion of what it is capable of is due to Creole, the PHP5 database abstraction layer. It’s essentially PEAR::DB on crack. Based on JDBC for Java, it takes an object oriented approach to interfacing with an RDBMS.

I can’t wait for PHP5 to get out of beta. There’s so much neat stuff waiting just around the corner. Hell, maybe I should give it a shot now and see just how stable it is and how well it will run my PHP4 code.

Inklog now uses Subversion (instead of CVS) for version tracking and control. ViewCVS (the old interface to view a CVS repository on the web) currently doesn’t work due to some quirk in the Debian Subversion libraries. However, the repository is visible on the web. Additionally, I’ve made an RSS 2.0 feed of the SVN commits thanks to a small XSLT translation.

It’s just my luck that when I finally decide to make the switch from CVS to Subversion, the Debian packages are having problems. Fortunately, it’s only with the Python bindings and not the whole of Subversion. So it just means that viewcvs won’t work until they get it fixed.