Last modified: 2013-10-08 22:04:59 UTC
When editing, clicking in the middle of or adjacent to an existing word in Latin, Hebrew, Arabic or Cyrillic scripts (others not tested) and pressing ctrl+k or clicking the link icon selects that word as the suggested sequence that is desired to be linked. When doing the same in Thai text nothing is selected by default.
Confirmed in Thai Wikipedia and in the English Wikipedia with Thai words. Furthermore, if you Ctrl-K on a Latin word and then Ctrl-K on a Thai word, the Latin word appears in the input box. See this test case: https://en.wikipedia.org/wiki/User:Amire80/VE-Thai
I tested with Telugu and Malayalam, and this doesn't happen there. Must be something special with Thai.
This is not a "bug" per se - VisualEditor's word boundary detection only works for scripts that use word boundaries. Consequently it doesn't operate on Chinese or Thai (which don't use word boundaries), but does for Korean, Latin and Cyrillic languages, and others (which do). Theoretically it would be possible to build some code for applying algorithmic word boundary rules for Thai, but that's both complex and slow, so if that is desired it should be opened as an enhancement (and we'd need to see how we can make it only slow down users who are editing in Thai). Marking as INVALID.
