If I have two words (picked at random from the Lao dictionary file, word break finds the expected boundary between them.
[is that of testdata/rbbitst.txt|format]
If I include some Lao digits (໑໑໑) between the two words, all boundaries disappear.
I noticed this behavior would happen while looking at the code. It doesn't smell right, I would expect some kind of boundaries to exist.
or some such.