Hi, There is a unicode character at 200B that is used for discretionary line-breaks (DLB) quite in the spirit of discretionary hyphens. In Tamil words broken at the end of a line are not hyphenated (as it is agglutinate language and is not a isolating language like English). Some editors like Emacs and InDesign do not allow cursor to move freely between characters, so I worked out a solution by putting these DLBs after every half consonants, which seems to be a nice solution to the hyphenation problem as well (but this may not be sufficient).
Wondering if XeTeX care about DLBs (they are useful to break long URLs and stuff like that). Suki
