Taco Hoekwater schrieb: > I may be missing something here, but couldn't you use opatgen > (a patlib frontend) and do the whole process in unicode? > > See http://www.fi.muni.cz/~xantos/patlib/
Are patterns generated by patlib superior to patgen (in terms of overall size, number of needed levels, etc.) or is it merely patgen implemented as a library? I took a glance at Antoš' thesis, and found > Design goal 7 Efficiency. The pattern generator—dealing with large > data—should be efficient. But we should not optimize the code paying > with concessions to the design. (page 20) and > [Thai patterns] We are able to cover 98.4% of word boundaries with > error 0.05% in mere four levels. (page 44), but no comparison to patgen. Best regards, Stephan Hennig
