[Mediawiki-l] mediawiki 1.5: so difficult?

[Agon S. Buchholz]

Rowan Collins wrote:

> 2) Can you clarify what you think makes the current customisation 
> process difficult, and - even more usefully - can you suggest how it 
> might be improved?

I think, what most people installing MediaWiki find confusing, is the 
absence of any centralized administrative interface; some parameters are 
configured in a script (LocalSettings.php), others directly in the 
database (e.g. Interwiki links), and again others in the 
MediaWiki-namespace etc.; additionally, there are some configurations 
seen on the Wikimedia sites which are completely undocumented, e.g. 
based on extensions, or MonoBook.{js|css} hacking, or templating etc., 
and a lot is changing between the releases (e.g. location of the 
translations for interface elements). Even if one was involved for some 
time in a Wikimedia project, one won't know exactly how all of this is 
tied together. Beyond that, what makes running your own MediaWiki site a 
real challenge is the current state of the documentation (e.g. 
explaining the Trackback functionality someone asked a few days ago, 
which somehow exists, but without any traces of documentation). A quite 
easy approach could be to create *one* single page explaining the 
concepts behind this ("why which feature is managed where and how"). 
Another, but probably unrealistic approach would be to write an 
administrative interfacee which should be task (and not technically) 
oriented.

Because of this, IMHO the documentation is crucial; a friendly usability 
(whith a task-oriented interface) would be better, of course, but the 
world isn't perfect anyway; with an accourate manual and working 
examples one can at least *try* to make things work; but if the 
documentation is out of date or incomplete, you have to understand the 
code to use the beast. And that's something beyond the scope of most 
people trying to set up a MediaWiki installation.

Since MediaWiki is Open Source and anyone could at least try to update 
the documentation, or draft some kind of administrative interface, there 
is not much to criticize but ourselves. However, having tried to start a 
complete and systematic rewrite of the documentation a few months ago 
myself, I learned that this can't be done in a convenient amount of 
time, not if the manual is supposed to be well-written, easy to 
understand, and should contain tested examples, and go deeply into 
discussion of important administrative aspects like security, 
replication, and high availability . Even if I'd write this im my native 
tongue (which would hopefully sound clearer than this mail), it would 
take approximately 9 to 12 months for a small group of people (1-2 
writers, 1 techy, 1 spell checker/editor), and require a considerable 
amount of research, and result in a 900-pages-book to be printed most 
favourably by O'Reilly ;)

The bottomline: The Wiki collaborative authoring priciple seems to work 
well for information which can be segmented in distinct particles (like 
Wikipedia articles). It seems to work less for information with some 
degree of required linearity (like a manual, or a textbook, which starts 
with the basics, finishes with advanved stuff, and has to be consistent, 
coherent, and cohesive in itself). Since it's improbable that one single 
person will do the task, the real challenge I see is how to utilize the 
Wiki priciple for the creation of a considerable larger and more complex 
document like the MediaWiki manual...

Just my thoughts,
Regards from Berlin, -asb