On Wed, Apr 28, 2010 at 19:24, Stephan Hennig wrote: > > Oh, please don't care about the timestamped patterns. The time stamp is > only a crude hack to avoid file name confusion while testing patterns that > are under development. We are happily waiting for Polyglossia to provide a > way for ageing patterns some day.
I wanted to say: timestamped patterns would need some extra handling anyway. LuaTeX would actually be a much better candidate for proper handling of as-many-patterns-as-one-would-want than pdfTeX. > The patterns in the dehyph-exptl package were originally meant for LaTeX and > Babel only. One question: do your patterns bring anything at this moment when one wants to use lualatex? That is: do we need to consider them now? > Mojca just grabbed our patterns for XeTeX (for converting the > ugly conventional German OT1/T1 patterns to UTF-8 seemed hard), Not hard, but if your patterns are better, there's no need to use the old patterns. The reason why we decided to load the old file (as opposed to integrating it into hyph-utf8) is that we didn't want to maintain two versions of German patterns in repository. Also, the OT1 support (parallel to T1 and duplicating the same patterns in two different encodings) is really really ugly. > arguing that > there's no need for (self-)backwards compatibility as long as XeTeX's > version number is lower than 1.0. So, as long as the latter holds, there is > no need to consider timestamped patterns in hyph-utf8. But once XeTeX or > LuaTeX grow v1.0, it would be nice to have an official solution for ageing > patterns in TeX. Freezing patterns is bad, IMHO. I agree with all of that. But ideas are welcome. If some elegant solution pops up, it would be nice to support it officially in hyph-utf8 (or maybe hyph-utf8-addons that would contain lots of different versions of patterns :) Mojca
