Last modified: 2012-12-30 20:59:18 UTC
Each category has its own sorting order (customizing of that is Bug 28397), and when a category is shown the collation is splitted up by starting letters (of the sort key). For big categories and categories sorted by a date it would be very useful to change the behavior of that splitting prozess. E.g. in a category where all sortkeys start with a year the sections should be titled with four digit numbers. I think the easiest approach would be have a magic function where you can declare how many digits should be used to split the item list. So {{#categorysections:4}} would make "Hello World" stand under "Hell" and "1984 (book)" under "1984". It also could be more flexible by defining in what ranges the rule should apply. E.g. {{#categorysections:4|1|9}} would make "1984 (book)" stand under "1984" and "984" stand under "984" cause they're in the range from 1* to 9*, but "Hello World" would stand under "H" as regular.
This would be difficult to implement for uca based collations (I believe anyways). For the "uppercase" and "identity" collations it would be more do-able to implement (not sure how it'd work with the magic we do for Hangul syllables though), but I'm not exactly sure how one would cleanly hook a parser func into Collation::getFirstLetter. This is very similar and possibly a dupe of bug 2802.