Page MenuHomePhabricator

Raw sorting in <languages /> gives unintuitive and seemingly random output
Closed, DuplicatePublic


In rETRAd3935a65cfca, ksort was introduced to sort languages by autonym. This is not only somewhat biased by placing languages using Latin letters to the beginning, but is also case sensitive, meaning that English is before català. I haven’t found any discussion about this change at all, neither at the Gerrit change, nor in the linked T64702 task. I suggest reverting this part of the change, especially as language code is the sorting order in sidebar as well for the vast majority of WMF sites, including all multilingual ones, so this list became less consistent with the sidebar one. Even if not reverted, the language names should be normalized before sorting, including converting to either upper or lower case, removing accents (e.g. čeština→cestina), resolving ligatures (Ænglisc→aenglisc) etc.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptSep 28 2018, 1:58 PM