On Sunday 11 June 2006 02:48, Morten Blaabjerg wrote:
This is a topic which interests me as well. I don't know of any forum other
than this list, in fact I find it a perfect spot for this kind of
discussion. Only if we know better what kind of directions we want wikis
and MediaWiki in particular to take - the better will the developers know
which kind of code solutions and architectures to pursue. So I welcome this
sort of discussion.
I will have a go at some points which I feel need to be adressed (and some,
which are currently being adressed).
Our wiki is used as a combination of knowledge repository, collaborative
works, articles, images and notes storage room, networking tool, news site
(sort of) and as an index of websites and projects, companies,
organizations and individuals - all related to the culture and media
landscape in the broadest sense.
That's more or less what we are doing. I work as a system
administrator/operator/help desk for a company that provides software for
online brokers as well as manages a number of datacenters for various
customers. We have about 20 different customers, each with their own
environments. A few are online brokers with two dozens machines in several
network segements spread across multiple physical locations, connection to a
dozen external systems,and so forth. At the other end are small Webshops with
a single machine and an extra cold-standby.
All told, we have over 1000 pieces of hardware that we manage. The asset
management is done with a help desk tool, but it isn't set up to store things
like configuration and troubleshooting information at the level we need it.
We have over 150 different applications/programs that we use in one way or
another. These rangein complexity from things like Oracle and MySQL to rsync.
However, we are all expected to know how to configure, manage and
troubleshoot each of these applications (plus the hardware).
Then we have all of the administrative information like how to open service
calls, who to notify at the customer's site when something happens, what to
pay attention to in our daily datacenter walk-throughs.
One obviously important aspect is find the information again. Currently we
store things in MS-Word Documents using Visual Source Safe. Independant of
problems storing information in monolithic documents like MS-Word, there is
the problem of checking the out of VSS, editing them and checking them back
in. Not to forget is the fact that only the department head and team leaders
can add documents to VSS. That makes adding new information to this pool
cumbersome at best. As a result people are less willing to store information
there and either have their own repository (mine was in HTML files) or they
simply don't write anything down.
It is currently only available in Danish - and one of
the next big
challenges will be to introduce new languages, english in an international
version in particular. I am following with great interest the project of
working-in an multi-lingual architecture into MediaWiki, using just one
database - as I feel this has the promise of simplyfying a great many
things. We don't want users to spread their activities over several wikis -
we want one user base to maintain the complete wiki over several languages,
across languages. The proposed solution with a single page being available
in several top language-namespaces sounds very promising. I wonder how far
off this project is?
My Wiki at home is in English and the one at work is in German. For us, there
is no need to have multiple languages. However, I also run the Linux Tutorial
(link below) and have been considering moving it to a wiki. I have been
working on the idea of having it available in multiple languages. One
important things for me would be to have those part which are not yet
translated into a specific language still appear in English. Or maybe a user
configuration option that switches this on/off.
One other thing on our wishlist, is better control
over categories and
namespaces. Adding categories to a page could be a lot easier. The same
goes for changing a page's namespace, and defining how category content is
displayed and sorted on category pages, other than the alphabetical order
(pages sorted after namespace, content, beginning paragraph, most popular
etc). An easier upload procedure, as has been suggested on the list
previously also would be nice.
It would be nice if the admin could define a set of categories which are then
available from a drop-down menu. Alternatively, you could have a set of links
on the edit page and clicking on them uses Javascript to insert the necessary
text into the form.
Personally, I good enough with cut-n-paste that I don't see it as a big
problem. However, currently anyone can create a new category and you
typically end up with a mess. For example, you have the categories
Application and Applications. With a menu the admin could ensure that the
name are consistant.
But this is just a wishlist. MediaWiki has only just
started and I am
confident that the software will only just improve and be even more
flexible and powerful in time. It already is a pretty potent piece of
software.
At this point I am not too worried about the flexibility since it is open
source and I am well-enough versed in PHP to make changes myself. :-)
However, I agree with you that it will improve as time goes by. Considering
how useful it already is, I am anxious to see how it is in a couple of years.
I would like to see is better seach capabilities. It would be nice to be able
to seach by category as well as a little more intelligence. For example, on
my personal wiki I have a page that described what directories are backed up
and how. So, I search for "backup" and nothing is found. I need to search for
"backups". (or is this a problem with my site?)
I have had a great time introducing the use of
categories for real on our
site, and we're only just beginning. I've made the Special:Categories page
into a sitemap, which is linked to from all categorized pages. Only problem
is that our top-level categories doesn't link to the new sitemap, since
they do not belong in any category. Of course, I could (and probably will)
simply solve this by creating a category of 'top-level categories' ;-)
The idea of a site map is not a bad idea. There is obviously a lot of manual
work to do (or not?), but that offers one more method of finding the
information. I have run into the argument that the information **has to** be
stored hierarchically. That there is some absolute requirement somewhere.
Granted humans natually put things into categories, but I see that most of
the belief that things must be in hierarchies is because that is the way we
are used to storing files on hard disks. Thus we store files for our web
servers like that. Also, up to now, most web sites have had some kind of
hierarchy due to limited search mechanisms.
I am not exactly sure what you mean by "top-level categories doesn't link to
the new sitemap, since they do not belong in any category." What would be a
top-level category on your site where you do not yet have wiki categories. I
am asking because we are working on a start page with various blocks for "top
level" categories. For example, administration, technical, customer, etc. To
some extent, these are too vague to put into wiki categories, so I am not
sure if it really makes sense to create wiki categories.
<snip>
What I would like to see is a hook that allows easy access to other data
sources. For example, as I mentioned we have a help desk tool which keep
tracks of assets, customer organizations and people, service contracts
(internal and external) and so forth. It would be nice to be able to pull
things directly out of the database. For example, on a given customer's
start page, there is a tag {{ServiceDeskCI:MaschineName}} that pulls specific
information for the configuration item (asset) or
{{ServiceDeskCustomer:CustomerName}} which pulls specific customer
information.
Another thing is printing of lots of articles. We have a requirement to
deliver a handbook to certain customers once a month (whether things have
changed or not). Since we currently use MS-Word, it is fairly straight
forward to create a PDF. Although I have written a script to convert the HTML
pages on the Linux Tutorial to a PDF and it would be fairly easy to create a
script that pulls pages out of the wiki, parses them and passes them to the
same mechanism, it would be nice if this was built in. There was a post on
the MW users forum from someone who creates "site books". To me, this would
be another good use of a process like this. Granted something like that is
not the goal of a wiki, but more like DocBook. However, DocBook does not have
nearly the same features as MediaWiki and I see it would be harder to make
DocBook fit for our needs like a Wiki as opposed to making a wiki do things
that DocBook does (note that the DocBook knowledge base is a wiki and does
not use DocBook itself).
regards,
jimmo
--
---------------------------------------
"Be more concerned with your character than with your reputation. Your
character is what you really are while your reputation is merely what others
think you are." -- John Wooden
---------------------------------------
Be sure to visit the Linux Tutorial:
http://www.linux-tutorial.info