migrating to mediawiki
Brian Rice
water at tunes.org
Wed Oct 19 22:42:26 PDT 2005
On Oct 19, 2005, at 8:47 PM, Massimo Dentico wrote:
>> Dear Tunespeople,
>>
>> are there people here with some energy for maintaining the Tunes
>> project? The Wiki is in a sorry state, and needs to be redone.
>>
>
> Hello Faré and Tunespeople,
>
> motivation and time lack: IMO wikis (at least wikis that I know of)
> are hopeless to maintain without *huge* manual intervention.
>
> For example, starting when CiteSeer passed from NEC to PSU[1]
> (with mirror at MIT[2] and UniZh[3]) all related links on our
> CTO[4] are in need of an update. So, we need to:
>
> a) locate each node with a link to CiteSeer;
> b) update such links, possibly agumented with a link to the
> same paper on mirrors.
>
> Without an account on tunes.org this is impossible to do
> automatically; even with such account, it is inconvenient
> and error prone (you know, regular expressions sometimes
> have some surprising, hard to notice, results on some texts).
>
> So, I ask: why not morphing CTO[4] in something really TUNES-like?
Hello? The only reason the wiki is locked down is because the
software can't handle spam. Mediawiki can handle spam. Link
translation is not hard - it can be handled. Even if we have to lose
a bunch of links and reviews, who cares? I don't even use CTO any
more because I can't stand signing in to fix it. Also it duplicates
rather poorly a lot of Wikipedia info. ALSO, LTU will be making a
wiki and handling the Citeseer issues for us.
> As Theodor H. Nelson put it, embedded markup is to be considered
> harmful[5]. There are some software which reified links and anchors,
> but they are, sadly, forgotten. See for example Atlas[6] and
> Hyper-G[7].
>
> When links and nodes are in a RDBMS, manipulating the graph
> structure is certainly simpler than editing text files manually
> or with text processing tools: to solve the links update problem
> previously stated, a simple SQL UPADATE statement suffice. Also,
> referential integrity is checked immediately when an INSERT or
> UPDATE is performed.
>
> I think that a quite robust and portable solution can be obtained
> starting from (ideas of, if not source code) Atlas[6], adapting
> it to PostgreSQL[8] (or MonetDB[9]) and interfacing all to a web
> server with FastCGI[10] (supporteed by most web servers, Apache
> included).
>
> Alternatively, has someone here some experience/knowledge about
> so-called Content Management Systems (CMSs)?
>
> Comments?
Oh. Sure, I absolutely agree. BUT WHO'S GOING TO CODE IT?
Slate doesn't exist for no reason - object-based multimedia is
supposed to be in there. Put your money where your mouth is, Massimo,
and CONTRIBUTE!
Notably, a CMS is one project that could be done with the CLIM-ish UI
project of Slate. I am not going to explain what a CMS is - look it
up. Squeak has half a one called SmallWiki2, but it is very
incomplete. The others in python and so forth are big and bulky for
language reasons. I have a friend who hosts Plone installations and
he hates it and always asks me when I'll have one in Slate ready.
My conclusion: help get some of this wikimedia migration done, or
write software, or shut up.
--
-Brian
More information about the TUNES
mailing list