With the use of `self`, even if the class name is changed (which is
very unlikely), then the code won't break. So it's a way of future
proofing the code.
Change-Id: Ib4046315b8aaee2e77af036893924962d25ceeb1
This removes the special collation for Northern Sami
that was added in 396007, when UCA support for Northern
Sami was not yet in MediaWiki. Now it is, so this
is no longer needed.
Bug: T182431
Change-Id: I760eb7ae8bf92f0ac93b5fca5cb69148a28d8f6f
This workaround was needed when ICU in production was broken
but after T189295 this is not needed anymore and we switched off
this collation from all Persian Wikis already
Bug: T139110
Change-Id: Ifad89555b6ac96a3eb36ca24b55e1f8ee57a1f05
* Adding new class AbkhazUppercaseCollation, mapped to 'uppercase-ab'.
* Extended CustomUppercaseCollation with support for sorting digraphs
and for alphabets larger than 64 letters (up to 4096).
Bug: T183430
Change-Id: I16d44568e44d7ef5b39c38b1a6257b9fe10a34d4
This commit adds a custom collation order for
Northern Sami ('se'). Northern Sami exists in ICU,
but the version of ICU that Wikimedia uses is a
few years old, and does *not* include Northern
Sami. It could be years before Wikimedia's production
servers use the one that includes Northern Sami (see
bug), so this is a temporary workaround to amend this
issue.
Bug: T181503
Change-Id: Ib8a48b8db99bef8ec4b05144aace5dbdcacfeded
This is based on a numeric uppercase collation. Bashkir characters
will be remapped to the private use area for the purpose of sorting.
Bug: T162823
Change-Id: I65f1af0b57ff6ded7d464e39efd401f178a3519e
We have several such collations defined in IcuCollation:
* bs-Cyrl
* de-AT@collation=phonebook
* fr-CA
* sr-Latn
They couldn't actually be used.
Change-Id: I3a62073583c49d3e90910aa8240fe9fcc0682386
This will cause the numeric collation to sort localized digits
for the current content language the same as how 0-9 are.
This only deals with the localized digit numbers, commas
and other number formatting are still not handled. Weird
"numerical" unicode characters are also not handled.
I was unsure if to make a "family" of numeric collations
where you specify numeric-<lang code>, or if it should
just use $wgContLang. Given that $wgContLang effectively
never changes, and also affects all other digit handling,
I opted to just use $wgContLang.
Any wikis currently using the 'numeric' collation will
have to have updateCollation.php --force run after this
change is deployed. At the moment that includes:
bnwiki, bnwikisource and hewiki
Bug: T148873
Change-Id: I9eda52a8a9752a91134d1118546b0a80d3980ccf
This collation orders text with numbers "naturally", so that
'Foo 1' < 'Foo 2' < 'Foo 12'.
Note that this only works in terms of sequences of digits, and the
behavior for decimal fractions or pretty-formatted numbers may be
unexpected.
This is only expected to work mostly correctly for English-language
text. Consider it a proof of concept. You probably want to use
an UCA collation with '-u-kn' suffix rather than this.
Bug: T8948
Change-Id: Ie268f2d92c5c75d0aaecf54ede2bdda1af3b309d
To use, add '-u-kn' to the end of a collation name and set it as
the value for $wgCategoryCollation.
Bug: T8948
Change-Id: Ica7908daf80624fa2648127114d01665e96234c0
