In the past few days, we have set up a few Squid servers on the newly
installed Yahoo! cluster in Seoul, South Korea. They have started
serving requests from several Asian countries last night, as can be seen
on http://noc.wikimedia.org/stats.php
(yellow = yahoo, blue = florida, green = amsterdam, purple = paris)
Work is underway to experiment with using the remaining boxes in the
Yahoo cluster as apache/mediawiki application servers, to offload the
Florida cluster and make things more redundant. As this will create many
complex problems and management overhead, we are not sure if and when
that situation will be used in production.
--
Mark
mark(a)nedworks.org
I would like to remind everyone that my citation feature is still
sitting in extensions/Citation.php waiting to be turned on on the
wiki(p|m)edia projects. I just fixed a nasty display bug and hereby
declare it "production ready" (M$-grade only, of course;-)
Remark: I tried to add a <note> tag using the same sytem, which seems
straightforward enough, only to discover that the parsed replaces custom
tags ordered by tag/position, not position/tag. So, if I were to write
citation/note/citation, it gets numbered 1/3/2 instead of 1/2/3.
Therefore, I did not implement <note>, everything will have to go
through <citation>, prefferably via template, which allows for
{{note|text}}. That will have to suffice for now :-(
Magnus
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
This was proposed at the village pump (
http://en.wikipedia.org/wiki/Wikipedia:Village_pump_%28technical%29#Correct…
) and I think it's a great idea. The idea is this: with a combination of
Javascript and server side tools, it should be feasible to double-click
any word on a page and then edit that single word. This is especially
useful for correcting spelling errors that you would be too lazy to wade
through an entire section to get to.
Considering how to determine what a word is may be a tricky task
though... is this feasible? How about enabling it for anonymous accounts?
- --
Edward Z. Yang Personal: edwardzyang(a)thewritingpot.com
SN:Ambush Commander Website: http://www.thewritingpot.com/
GPGKey:0x869C48DA http://www.thewritingpot.com/gpgpubkey.asc
3FA8 E9A9 7385 B691 A6FC B3CB A933 BE7D 869C 48DA
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)
iD8DBQFDJ3pkqTO+fYacSNoRAsyFAKCFM0xcIP3sGV1quhXsSllaeW/hxgCfUr4B
Xsz3ltcSgp6dtVx73yAoPAo=
=flK4
-----END PGP SIGNATURE-----
What is/are the reason/s for storing the full text of page revisions in
the database as opposed to some form of differential? Am I correct in
assuming that speed has been given priority over storage space
requirements, and if so, has any benchmarking been done to find out how
much overhead would be added by storing revision as diffs and how much
space would be saved?
Also, has there been any discussion of the possibility of branching a page
(as is possible in e.g. a CVS repository)?
I haven't seen anything on either of these issues after a quick search
through the list archives and some non-exhaustive reading of the main
wiki, so pointers to anything I've missed are welcome.
--
http://members.dodo.com.au/~netocrat
Today I decided to analyze in more detail to what extent articles across
Wikipedias remain protected for long periods of time. The report is at:
http://meta.wikimedia.org/wiki/Longest_page_protections%2C_September_2005
(To developers: The script I used is 'logprot.pl' in my home directory.
It may be desirable to make this available as a special page, if someone
can figure out a way to make the query scale.)
It shows all pages in all language Wikipedias that have been protected
for more than 14 days. Note that, by the time you look at it, some of
the pages in it may have been unprotected already.
The Wikipedias with the most such protected pages are (article rank in
parentheses):
German - 253 (2)
Japanese - 165 (4)
English - 138 (1)
Italian - 19 (5)
French - 15 (3)
Spanish - 13 (10)
This confirms my intuition that long term page protection is used
excessively on the German Wikipedia. It is quite striking that many,
many controversial articles have been protected for months. For example,
articles about veganism, sex, democracy, abortion, astrology, Karlheinz
Deschner (famous atheist writer), Silvio Gesell (controversial
economist) and his Freiwirtschaft theory, Gorleben (controversial
nuclear waste disposal site), and Egon Krenz (East German politician)
have been protected since July. Articles about child sexual abuse and
pedophilia have been protected since April 2005 and March 2005,
respectively. Notably, in the child sexual abuse case, the article was
also cut down from 54,000 characters to 2,000 before being protected,
making it effectively useless -- a rather drastic measure to deal with
ongoing controversies.
The longest protected articles appear to be related to German student
corporations. The record holder is [[de:Schmiss]], which has been
protected since January after a neutrality dispute.
Perhaps ironically, even the article about Wikipedia itself has been
protected since August 25.
Note that the local policy on protection, at
http://de.wikipedia.org/wiki/Wikipedia:Seitensperrung and
http://de.wikipedia.org/wiki/Wikipedia:Administratoren , is not unusal
and recommends only short protections except for very high exposure
pages like the Main Page, or unimportant pages like redirects which are
frequently vandalized. This raises the question why no bold admin has
unprotected these articles yet.
I cannot say anything about the protection patterns on the Japanese
Wikipedia, which is the only one which stands out besides English and
German. The long term protections on the English Wikipedia appear to be
mostly accidental. When someone notices that a page has been protected
for very long, it is generally quickly unprotected.
Across languages, possibly with the exception of Japanese, the German
Wikipedia is alone in the pattern of locking down controversial articles
for months. Protected articles also seem to not be tagged as such, so
that visitors do not see a reason for the protection on the page (a
visible marker might also encourage sysops to unprotect the page).
One immediate effect, besides stagnation, is that sysops become far more
relevant in the power structure, as they are the only ones who can add
information to articles after protection. Instead of being janitors,
they become editors. This, I believe, must have social repercussions
beyond the articles concerned.
I can see three immediate ways to address the issue, by increasing
complexity:
* limit protections by policy
* add an automated or template-based visible marker to protections in
the article namespace
* add an "expiry" feature for page protection similar to blocks
I am merely reporting this issue and will leave it to others to deal with.
Best,
Erik
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
> I see I'm not the only crazy guy who uses PHP for a bot. Good to know
> that.
I'm currently developing a PHP bot for Wikipedia too (see
http://en.wikipedia.org/wiki/User:Ambush_Commander/LinkFix_dump ).
> Anyway, try the Advanced HTTP Client
> (http://www.phpclasses.org/browse/package/576.html).
I use SimpleTest's simple browser for web browsing (
http://www.lastcraft.com/ ). No point in reinventing the Web retrieval
wheel.
- --
Edward Z. Yang Personal: edwardzyang(a)thewritingpot.com
SN:Ambush Commander Website: http://www.thewritingpot.com/
GPGKey:0x869C48DA http://www.thewritingpot.com/gpgpubkey.asc
3FA8 E9A9 7385 B691 A6FC B3CB A933 BE7D 869C 48DA
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)
iD8DBQFDKKx7qTO+fYacSNoRAtUsAJ97vZK3X4+Q0Tv08+86i5Dlz79S1ACfTcov
OHJ0oKYmiJTcdaDgyBW+dbc=
=ldFu
-----END PGP SIGNATURE-----
Quick (or not so) question;
I want to add a link to create a new article from the search page when
there are no matches returned for the search, i've gone over and
echoed what is stated in
http://www.wikicities.com/wiki/Help:Searching#Create_new_articles_from_the_…
but I get nothing on the search page after, anybody have any ideas?
thanks!
Hi everybody!
As most of you probably know there are quite a few
requests for Wikipedias in new languages to be
created. Many of them are still discussed
controversially. Others are not but are unequivocally
welcomed. Nevertheless, no new Wikipedias have been
set up for about three months now and we might lose
potential new contributors if we leave them on
stand-by for too long.
The most clear-cut case seems to be the Neapolitan
Wikipedia. Matter of fact, a large number of
Wikipedians are supporting the idea, noboby opposes it
and several native speakers are willing to contribute
and waiting to get under way.
I know you guys are extremely busy. Could someone
please find the time to take of this and install a new
wiki for Neapolitan? I'd love to do this myself but
unfortunatly can't.
Details are here:
http://meta.wikimedia.org/wiki/Approved_requests_for_new_languages
Regards,
Arbeo
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
Hi!
Can anybody tell my what differencing algorithm MediaWiki uses?
DifferenceEngine.php starts with "see diff.doc" - a file I cannot find.
As far as I know it has been taken from phpwiki but is it still the same
algorithm and what algorithm does phpwiki exactely use? I have not found
any better references on differencing algorithms but the papers listed
at http://en.wikipedia.org/wiki/Diff#References. There seems to be some
new research in differencing XML-files. Is MedaWiki's differencing
algorithm based on E. Myers's algorithm (1986)?
Thanks and greetings,
Jakob