Page MenuHomePhabricator

Case sensitivity problem with GlobalUsage
Open, MediumPublic

Description

It seems that there are some problems with the Global Usage extension and wikis with case sensitive initial letters (like Wiktionaries).

See http://commons.wikimedia.org/w/index.php?title=Commons:Village_pump&oldid=40453079#Incomplete_usage_listings

When the Wiktionaries went case-sensitive, the file usage checking from Commons became case-sensitive. However, the case-sensitivity on Wiktionaries applies only internally, and does not apply when an external link is made to another project. In particular, links from a Wiktionary to files on Commons are not case-sensitive because the link is made to a case-insensitive project.

As a result, Commons file usage information only displays some of the instances of a file used on the Wiktionaries. For example, File:La-cls-discipulus.ogg is used on the English Wiktionary, but its page on Commons says "There are no pages that link to this file."

Commons should realize that links made to it from *any* project are de facto case-insensitive because it is itself case-insensitive.


Version: unspecified
Severity: normal

Details

Reference
bz23947

Related Objects

Event Timeline

• bzimport raised the priority of this task from to Medium.Nov 21 2014, 11:04 PM
• bzimport added a project: GlobalUsage.
• bzimport set Reference to bz23947.
• bzimport added a subscriber: Unknown Object (MLST).

Bryan.TongMinh wrote:

This can either be fixed at insertion, by always capitalizing the input, or at query, by querying both the uncapitalized and capitalized version.

The first one gives a problem if both capitalized and uncapitalized shared repos are used, because the capitalization would need to be fixed when a file is deleted on one of the shared repos and exists on the other.

The second one is just nasty, but I'm still going for that one.

Bryan.TongMinh wrote:

Patch that implements option 2

Patch; needs testing; posting here so I don't lose it.

Attached:

*Bulk BZ Change: +Patch to open bugs with patches attached that are missing the keyword*

  • Bug 29803 has been marked as a duplicate of this bug. ***

baltoslavic wrote:

Bulk patch has not corrected problem on "duplicate" Bug 29803. See details there for continued problem. Links TO Wiktionary are case-sensitive, which is fine, but link searches FROM Wiktionary are also made case-sentitive for purposes of checking usage in Commons, and that is a problem. The search fails to find file usage, as is.

Not sure why the other bug was marked "duplicate", since the problem here seems to have been fixed, but the problem over there has not.

baltoslavic wrote:

For the record, this problem has now been around for TWO YEARS without being fixed. Is anyone working on it?

baltoslavic: See "Assigned To: Nobody" above, so nobody is working on this. To get things going, anybody could take the patch in comment 2 and test and improve it on a local test installation of MediaWiki or on a Wikimedia Labs instance.

Bryan: Any plans to pick up your patch again?