Hello vic, vic wrote: MediaWiki is originaly a hack of usedmodwiki to fits the need of wikipedia, it eventually got used by other people for different matters. Developpers are also somehow system administrators for wikimedia server farms. That's a call for new features, 1.4 have been under development for like 6 months and the .plan is to try to get it before the end of the year. I personally think we have to freeze new features right now and start beta testing. If we do so, new features will be included in january with a possible release (ex: 1.5 around april 2005). The good point is there is already some patches floating around: _ Enotif for email notification but it isn't in cvs yet. It looks complete and probably can be included in 1.4. _ LDAP authentication, the patch is small and available in bugzilla. I did not have the time to test it yet but it will certainly be included in 1.4 (relativly minor patch). Another patch I personally made is a new system to manage users, it's not intend to wikipedia but a common request coming from the MediaWiki user base. Also, I am not sure we want email notification in wikipedia, cause it also means the system team will have to manage two email servers and that will raise our bandwith usage. Anyway it is a great feature and deserves to be included in MediaWiki. Na no more branch, it's complicate the developper work. Actually our branches are: * 1_3 : used to release current stable releases (1.3.x) * 1_3A : tweaked 1_3 for wikipedia * HEAD : the future 1.4.x * SCHEMA_WORKS : a massive patch to change the database layout, will probably be 1_5 once HEAD is moved to 1_4. That also mean each time something is patched in HEAD for 1.3 it needs to be backported to the 3 others branches. In conclusion, MediaWiki development is a bit slow but it's moving. If you know about PHP and are interested in open source development you are welcome to join the team of developpers :o) cheers, -- Ashar Voultoiz - WP++++ http://en.wikipedia.org/wiki/User:Hashar Servers in trouble ? noc (at) wikimedia (dot) org

vic wrote: Development for Wikimedia projects is excessively slowed down by the constant need to cater for non-Wikimedia users. A slow release cycle suits non-Wikimedia users, because they want the software to work out of the box. Security flaws exposed in a release can't be rapidly corrected, because it requires hundreds of users to upgrade their copy. Wikimedia projects on the other hand would benefit from continuous introduction of features ported from the unstable branch, once a suitable testing sequence has been completed. Bugs, security flaws and other non-ideal behaviour can be easily corrected, since there is only one live copy to update. Currently, desparately needed features are delayed for many months because despite being tested, they are considered too unstable for release in a minor version increment. Readying the unstable branch itself for release requires many hours of work fixing half-completed features written by other people, committed and then abandoned by developers who have more interesting things to do. A methodology of continuous release to Wikimedia would at least provide an incentive for these developers to finish what they started. In my opinion, non-Wikimedia users have no right to expect MediaWiki development to cater for them. Their needs are incidental. So I agree with the OP, on a practical level if not philosophically. Another branch would suit me just fine. -- Tim Starling