MediaWiki – DailyStuff

Etch, vaarwel

Na meer dan twee jaar gebruik is er een einde gekomen aan Debian, de Etch release om precies te zijn. Recentelijk is ook de laatste machine geupgrade van Etch naar Lenny welke eerder dit jaar uitkwam. Het is nu wachten totdat de Squeeze release komt en de verwachting is in de eerste helft van 2010.

Tot die tijd is het kijken wat voor mijn mist in Squeeze, want voorlopig zijn lijkt Squeeze al redelijk te voldoen als backend. Helaas is er nog voldoende te doen aan de frontend. De frontend als in webapplicaties zoals WordPress, Mediawiki en Drupal. Dus de komende maanden maar eens kijken wat er nog mist in dat gebied en patches insturen.

Wiki spam, the new threat?

Web 2.0 would enable users to participate and turn the web into a place where users is at the center. But after a couple of years this maybe a dream as more and more robots come online and start to demand their place on the web. A place to help their masters who need to host backlinks, content or some weird project like the Graffiti Network research project at Brown University.

Looking at the logfiles of one Web 2.0 application based on MediaWiki gives enough to think about. The MediaWiki installation is protected by some extensions to prevent spam and this is the reason I didn’t notice some issues. Those extensions don’t stop abusers from trying to post (between 17 and 21 KB per post) their content to the Main_Page. And with about 10000 postings by 3600 machines for every day of the week it gives an indication how big the issue has become.

For now they are unable to post, but the network appears to be big and growing. And some day they will find a way to work around the current antispam extensions. So the time has come to start collecting data about those abuse attempts and match them with spam on weblogs and spammy e-mails. Hopefully there is a relation, but I fear the moment that Web 2.0 is realy to open for the world to handle.

First thoughts about WordPress 2.8

Going mainstream has its price and so has going with WordPress. It is a populair application and does work great as a drop-in-place-and-run application, but it kicks and screems when you try to tame it for mass shared hosting. And I still wonder who kicks and screems harder. The code or the developers.

For now 2.8 is on hold for me to see how I can modify it so it becomes usable again in a shared hosting environment. And the following line in the release announcement worries me.

The core and plugin updaters in previous versions of WordPress have been such a success we decided to bring the same to themes. You can now browse the entire theme directory and install a theme with one click from the comfort of your WordPress dashboard.

It is going to be interesting when I load it on a test container next week. Maybe I will update MediaWiki 1.14 to 1.15 first to get in a good mood.

SQL is not a standard?

SQL has been seen by many as a standard and on paper they are right. In the real world they are far from the truth when you try to make applications work on multiple databases. Who doesn’t remember ODBC as the golden bullet to solve all your database access issues and later on the same with JDBC. Luckily the language has been standardized is 1986 for the first time and they now are working on the 2008 revision which can be bought from ISO if you want to implement this free standard.

But who implements this standard? MySQL, Oracle, Sybase, PostgreSQL, MS-SQL? The question may be more like “who implements what?” and “how?”. Bug 18078 may give a hint in how well vendors are implementing SQL and may give an inside on how big the vendor lock-in really is. But is also gives an inside on how developers are wasting time writing and discussing abstraction layers to let there application like MediaWiki for example run on multiple databases.

Is this the new barrier where the FOSS-community needs to spend time to give proprietary vendors a run for there money? Just like Mozilla pushed Microsoft to accept open standard for the web, or like OASIS did with OpenDocument, or like the XMPP Standards Foundation is doing with instant messaging? Yes, AOL is running to get there AIM/ICQ-network migrated to XMPP so they can compete and communicate with Google Talk. Hopefully time will teach us how we can free us from proprietary only solutions and level the field again. Until then it’s something to work on and check for when using new applications.

MySQLism slaat weer toe

Hoewel de core van MediaWiki zelf redelijk goed met verschillende databases om lijkt te gaan is dit niet het geval voor sommige extensies. Zo ook voor de extensie NewestPages waar wordt uitgegaan van MySQL als database en de SQL-query zo geschreven is om op MySQL te draaien.

Helaas is PostgreSQL wat kieskeuriger en klaagt over het feit dat er een “SELECT … LIMIT 0,5” wordt aangeboden. Gelukkig is er ook een optie om met een SQL-statement beide database te bevragen. De vraag is dan ook waarom niet direct voor “SELECT … LIMIT 5 OFFSET 0” is gekozen aangezien zowel MySQL 5 als PostgreSQL 8 dit ondersteunen.

De developer heeft een bugreport en patch gekregen om dit structureel op te lossen. Helaas gaat deze functionaliteit niet zonder flinke aanpassingen werken op Oracle aangezien ondersteuning voor oa LIMIT en OFFSET daarin niet aanwezig is. De komende periode maar eens kijken naar welke extensies ook problemen hebben met PostgreSQL als database achter MediaWiki, want ik ga niet meer terug naar MySQL.