Le 27/04/2010 02:30, Karl Berry a écrit :
Yes on all counts. I+Elie+Khaled have already spoken to Peter B
(etex.src author/maintainer) about this. He is ok in principle with
integrating luatex changes into the master etex.src, but (as I
understood it) prefers that they be released as a separate
luatex-specific file first, so they can be more thoroughly tested before
the master is changed.
Great. I agree that it's better to get some testing in a separate file first,
but it's that Peter is ok with merging it later, so that we avoid having too
many copies of essentially the same code in the long run.
So our default plan here should be to ship an etex.src in the luatex
tree. Hmm, seems to me now that it doesn't have to be named etex.src,
since it's used at format time rather than buried in unchangeable TeX
macros. Right? If the luatex version can have a different name
(luaetex.tex comes to mind), that would be nice.
luaetex.tex sounds fine to me.
So, unless I'm mistaken, the remaining points to be sorted are:
- import the necessary files in tex-hyphen's SVN
- decide on a location for the new files in TL (should not be too difficult)
- sort out whether we can work our way in Lua with current language.(dat|def)
(would be easier from the TL infra POV, but require more work from the
macros/Lua POV) or really need an additional language.(dat|def).lua, which would
need adaptations in the TL infrastructure.
- get names for US english patterns work as Karl suggested
- it is not clear to me right now whether an update of luatex's specific
hyphen.cfg will correctly trigger a rebuild of the formats using it. (Well,
right now it isn't clear to me either that an update of hyph-utf8 triggers a
rebuild of most formats, so I'd probably better get some sleep.)
- do as much testing as we can
Am I missing something? If not, it sound rather reasonable to hope getting it
ready within a few weeks.
Manuel.
