Maha told from the last Wikipedia workshop and get-together in Cologne. It must have been a great success. People were exchanging efficiently and in a good atmosphere ideas, tipps and tricks how to do certain things in Wikimedia projects and in the underlying MediaWiki software in general and did create quite some new content.
Blackwing however couldn’t resist making some remarks that the sun is not always shining so bright and started with a larger slightly heated debate about his recent Wikipedia experience. His view was quite the contrary.
He was basically complaining that Wikipedia has such a great potential but that the underlying technology does encumber Wikipedia and Wikimedia as a whole. He was especially refering to server outages and that there are in fact people out there that have lots of knowledge on database management (the main Wikimedia performance bottleneck are the database servers) and who would volunteer and seem to know quite in detail how to speed up the database structure. He named a few issues like the current gzipped storage in MySQL and several ideas how to reduce the number of database calls in order to generate a web page a user is browsing (the red and blue links recursion problem for example). He said that that he and several others did speak to Wikimedia developers but did not recieve much positive feedback out of several reasons and were thus quite frustrated. He also told that there are people that even have just done a restructuring of Wikipedia database dumps and wrote a HTML output engine for it, for example the people of the German language read-only Wikpedia mirror lexikon.freenet.de (I don’t like the user interface there as it is IMHO really ugly embedded into the main freenet design but it is indeed quite fast).
So my conclusion is: From time to time we should just should focus on our own small world and try to do the things we want to get done on our own or in small groups and show in the end our results to the others. So I’d call to the people that have lots of knowledge on databases: Don’t try to persuade Wikimedia developers now although you’re maybe 100% right. Create a nice open project read-only Wikipedia mirror (I know answers.com but it is English language only and not what I am thinking about) and make it damn fast and if you’re done we have two things:
And of course if you have something that just rocks and which you can show to others it is much easier to convince people.