Last modified: 2008-07-11 18:13:46 UTC
When given a non-ASCII filler, padright: and its kin apply the right number of an incorrect element (see [[:meta:User:Taragui#Padright test]] for an example. Handling of Unicode seems broken in any case: non-ASCII characters in the first argument are counted '''according to their byte length''' (i.e., as 2 to 4 characters) instead of as one each, as they should. This breaks the fix for the unavailability of a <code>strlen</code> function proposed at [[:meta:Talk:ParserFunctions#strlen & substr]].
This doesn't break that fix, because that probably would have done byte count too. Note that Unicode characters can also be visually less than one character, e.g., combining or zero-width characters (although admittedly those are rarer).
Language::pad use strlen() whereas we should use mb_strlen() (and code a function if mbstring is not loaded)
Fixed the pad functions in r37567. Were there any others?
