Page MenuHomePhabricator

Category sorting should be case-sensitive on iswiktionary
Closed, ResolvedPublic

Description

Author: spacebirdy

Description:
Hello,
please enable differentiation between upper and lower case initial letters in categories again on is.wiktionary.
I don't know why this was turned off, it is essential for our categorization because in wiktionary we do differ between upper and lower case, so please turn it on again on is.wikt.
now all are mixed together, which was not the case before:
e.g.:
http://is.wiktionary.org/wiki/Flokkur:Nafnor%C3%B0

thanks.


Version: unspecified
Severity: normal

Details

Reference
bz30722

Event Timeline

bzimport raised the priority of this task from to Medium.Nov 21 2014, 11:55 PM
bzimport set Reference to bz30722.
bzimport added a subscriber: Unknown Object (MLST).

This has been the case ever since the new category collation system was introduced, which was in March IIRC. Changing summary.

To confirm -- this requests a change to iswiktionary (is.wiktionary.org), not iswiki (is.wikipedia.org)?

rm keyword shell for now - MediaWiki currently doesn't have support for this so its not a simple shell change. What needs to be done is one of:
*New "identity" collation that just sorts things based on code point. (This is very trivial to do)
*Or the locale specific collation stuff needs to be fixed up to use different locales (bug 30673)

I added support for the pre-1.17 behaviour in r96765. Re-adding shell keyword.

What needs to be done to fix this now is:
*(obviously) deploy the relevant revision
*set $wgCategoryCollation = 'identity'; for iswiktionary
*run the updateCollation.php script.

As an aside, a new collation could be devised for the wiktionaries that differentiated between upper and lowercase, well sorting in order (aka A followed by a followed by B followed by b...) which might suit them better than plain code-point order, but that'd be a fair bit more work.

As an aside (that's kind of off-topic), jbowiki should possibly have this too since they have $wgCapitalLinks set to false, and their language doesn't use capital links (so the first letter headers in category should probably be lowercase).

spacebirdy wrote:

(In reply to comment #2)

To confirm -- this requests a change to iswiktionary (is.wiktionary.org), not
iswiki (is.wikipedia.org)?

thanks for response.

only is.wiktionary, because there the entries differ between upper and lower case, so should the categories...

Closing LATER, needs to be done post 1.19 deployment (tracking bug added)

(In reply to comment #7)

Closing LATER, needs to be done post 1.19 deployment (tracking bug added)

Umm why? Code was merged into 1.18wmf1.

(In reply to comment #8)

(In reply to comment #7)

Closing LATER, needs to be done post 1.19 deployment (tracking bug added)

Umm why? Code was merged into 1.18wmf1.

orly?

Orly?

I ended up tagging it 1.18, and then Roan merged it into 1.18wmf1 in r97635

spacebirdy wrote:

Hello, when will this be active, it is still not working.
Thanks in advance.

(In reply to comment #11)

Hello, when will this be active, it is still not working.
Thanks in advance.

Oh...

Re-opening.

To shell folks: What needs to be done:
*$wgCategoryCollation = 'identity' on iswiktionary
*Run updateCollation.php on iswiktionary

14:00 <+logmsgbot> !log dzahn synchronized ./wmf-config/InitialiseSettings.php

and:

mwscript updateCollation.php --wiki=iswiktionary
...
80915 rows processed

done