After hacking it for so many years, there are still some surprising nuggets hidden in TYPO3. Our TYPO3-newbie webmaster Johannes recently pointed me to the RECORDS type in our template which is extremely useful for including portlet-style content elements in your template. In order to include a plugin somewhere in your page template, simply add it to a sysfolder or hidden page somewhere and refer to the content record like this:

subparts.TAGCLOUD = RECORDS
subparts.TAGCLOUD {
tables = tt_content
source = 444
dontCheckPid = 1
}

That’s it. The plugin content is rendered without USER_INT fiddling or COA tricks, configuration is dummy proof with Flexforms which seem to be more popular than TS configuration anyway. You can also include link lists, search forms, user login or normal content etc. like this, all nicely editable by your average users.

I don’t know why I only looked at Textpattern now but it surely looks nice for smaller sites that I set up for friends. I used to do everything with TYPO3 but had a hard time trimming it down for simple use cases with only a couple of pages and no dynamic stuff. Textpattern has a very intuitive templating system and comes with all the categorization and syndication features you might want for a start. The admin interface is ugly but so is TYPO3’s. And Textile is better than HTMLArea although not as cool as Markdown. I’ll give TXP a try for the next project.

The underused webserver has just gotten another resident. Welcome back to the blogosphere, Steffen!

Today Robert and Sebastian announced that TYPO3 5.0 will work on top of a JSR-170 compliant content repository, which, of course, has to be implemented in PHP first [sic!]. What does this mean? Clearly a landmark in TYPO3 history, as we are now officially embracing the J2EE specificationism that is elsewhere rightfully ignored. Instead of becoming more agile, 5.0 might be even more of a 500lbs gorilla than ever before.

This does hardly come as a surprise: If you’ve witnessed the discussions at the T3DD, there has been a fairly large Java/Enterprise crowd who constantly argued a) for making anything exchangable (components!), b) that content must be stored in RDBMS, XML, flat files and preferably punch cards (just in case), c) that we must support standards (lots of them) and d) TYPO3 must be designed for any purpose. In short: 5.0 should be a kitchen sink that might also be used as a CMS.

So what does the 300-page spec of JSR-170 tell us:

1. All data is stored in a tree, with nodes and properties (oh, really?!)
2. You can have any persistence backend (see below)
3. You can retrieve data via XPATH or (optionally) SQL-like queries

I’m not going to rant about the fun of mixing JAVA and PHP, which is required until we have a PHP implementation, or the man-months that need to be spent on the core data implementation (we’re not even talking about the actual content schema that is needed for TYPO3 as a CMS).

Ignoring the API for a moment, let’s have a look at the JackRabbit implementations for the persistence layer (i.e. our database):

Three out of four mature persistance managers use binary serialization for storage, which is great for transparency and enables us to use 0 (zero) of the industry-tested enterprise tools, like  RDBMS or file systems. The only other semi-mature option is XML-based storage, which is (surprise!) slow and even less reliable.

Please notice that those are all implemented by JAVA people who already have experience with JSR-170. In the TYPO3 community, as represented by the people at T3DD, even a simple ORM solution seemed to be quite a new idea to a lot of them.

I can only hope that the 5.0 project will not get completely off the trail by re-implementing, among others, ZOPE and Websphere in PHP while TYPO3 is still not less and not more than a CMS framework.

[The announcement was made on typo3.projects.typo3-5_0.general which is only available on the TYPO3 usenet (news://news.netfielders.de)]

T3DD06 was definitely the best conf I’ve been to, and I had a lot of fun meeting all the smart guys from -core and the community. The new TER frontend has finally been deployed and even the most annoying TIMTAB bug was fixed last weekend.

Other news … my thesis has been somewhat neglected lately but I did insane amounts of computation and data presentation last week. In a comparison of 25 countries any small table with 5 regression coefficients takes up a whole page. Not to mention the fun of presenting a dozen goodness-of-fit measures for every model and every country. Cross-national research surely does suck at times.