I've looked at this a bit more. There are more serious problems.
Apparently, no-one converted the 5.0 titles in the wiki to 5.1 when "normalization" was turned on; there are pages that can't be accessed. (!) for example, try this (Malayalam for "fish"):
http://ml.wiktionary.org/wiki/%E0%B4%AE%E0%B5%80%E0%B4%A8%E0%B5%8D%E2%80%8D
that gets normalized to 5.1, which is a redirect to the 5.0 form (in this case) which is normalized back to 5.1. (there is a variation in the 5.0 form too that complicates it) The content page exists (I can see it in the XML dump), but can't be accessed because there is no way of referring to it.
Was it necessary to force the normalization to 5.1? I would think just using the 5.1 forms by convention would be/would have been entirely adequate? Maybe with a bit of bot conversion? (Moving 5.0 to 5.1 leaving redirects, converting text while leaving iwiki links alone.)
The present state apparently can't be bot-fixed, as (some) content pages can't be read.
As it is, it is impossible to write valid iwiki language links to 5.0 forms on other wikis. One could create 5.1 redirects on the other wikis and link to them; but that doesn't help cases like above where one can't even access the content page. There are 998 of them (as of the last XML dump, 3 April) in this state apparently.
Mind you, I'm not sure I have all the details right yet, and I'd like to read through a current dump, now that they are running again.
Robert
On Sat, May 22, 2010 at 2:57 PM, Platonides Platonides@gmail.com wrote:
We should probably normalise to 5.1 on all wikis. I can view the 5.0 characters but not the 5.1 ones, though.
But would someone tell me where in the server code this is done? I have not been able to find it. Then I can understand a bit better, possibly just fix it in the bot code somehow, or suggest a fix server-side.
http://svn.wikimedia.org/viewvc/mediawiki/trunk/phase3/languages/classes/Lan...
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l