Community Wishlist Survey 2020/Archive/Choice of the wiktionary project-wide sorting policy for categories

From Meta, a Wikimedia project coordination wiki

Choice of the wiktionary project-wide sorting policy for categories

NoN Proposes a social/policy change rather than a technical feature

  • Problem: Several sorting policies can be used in dictionaries. The most usual one (policy 1) sorts according to letters and digits, removing everything else (e.g. red fox is considered as redfox for the purpose of sorting, and A/N as AN. However, when all phrases get separate entries, it may be better to use a different policy (policy 2): some dictionaries, including fr.wikt, consider that everything not a letter or a digit is replaced by a single space, except apostrophs, which are removed. Examples: presqu'île is sorted as presquile but A/N is sorted as A N and boulanger-pâtissier is sorted as boulanger patissier. This leads to an order such as boulanger, boulanger-pâtissier, boulangerie rather than boulanger, boulangerie, boulanger-pâtissier. The current fr.wikt solution is the addition of a sortkey to each page with a title including special characters such as - or ' or / etc.
  • Who would benefit: all editors creating pages for words with a - or ' or / etc.
  • Proposed solution: a project-wide parameter specifying one of the policies above (policy 1 or policy 2), or no policy ("no policy" would lead to the use of the category collation without any special treatment (assuming that the " Multiple collations per site" proposal is adopted, I hope so)).
  • More comments:
  • Phabricator tickets:
  • Proposer: Lmaltier (talk) 20:41, 8 November 2019 (UTC)[reply]

Discussion

  • Is this project-wide or should this alphabetical orders be recorded in MediaWiki in order to be use the same way in each Wiktionary and in each other projects using MediaWiki software out of the Wikimedia Foundation's projects? To phrase it from another point of view: is French tradition of ordering English the same as German tradition of ordering English and so on? Noé (talk) 14:15, 15 November 2019 (UTC)[reply]
    • It's a dictionary policy depending mainly on the number of phrasal entries in the dictionary; when there are many phrases as separate entries, one option is better; when phrases are included in separate word entries, the issue is different. Lmaltier (talk) 21:21, 21 November 2019 (UTC)[reply]
  • We have discussed this as a team and it seems there are no engineering resources needed. Simply find consensus for the collation you wish to have on your wiki, and request it on Phabricator. Please let us know if we're misunderstanding you. Thanks for participating in the survey! MusikAnimal (WMF) (talk) 17:37, 15 November 2019 (UTC)[reply]
    • @MusikAnimal (WMF): Simply find consensus for the collation: this proposal has nothing to do with collations (for collations, I support the "Multiple collations per site" proposal, as the collation needed depends on the category language). There are many possible collations (more or less one per language) but only a few general sorting policies used by dictionaries, the ones I mention above. Please read the proposal again: it's something very important in dictionaries. Rejecting the proposal without any discussion because of a misunderstanding would be a pity. Lmaltier (talk) 21:22, 21 November 2019 (UTC)[reply]