Thanks. "Fixed" at commit 0275521 (or attached ltj-base.lua.diff).
By the way, in my environment (LuaTeX rev 4917),
produces
> \y=macro: ->2.0pt minus 1.0pt.which should be 2.0pt minus 1.0fil, even the above patch is applied.
But I suspect that there is a bug of LuaTeX, since
%#!luatex
\directlua{
local s = node.new( node.id('glue_spec') )
s.width = 65536
s.stretch_order, s.stretch = 1, 10000
s.shrink_order, s.shrink = 2, 655360
tex.setskip(0, s)
}
\message{\the\skip0}
\bye
produces
This is LuaTeX, Version beta-0.79.0 restricted \write18 enabled. (./test.tex 1.0pt plus 0.15259fi minus 10.0pt ) No pages of output. Transcript written on test.log.which should be 1.0pt plus 0.15259fi minus 10.0fil.
h7k への返信
By the way, in my environment (LuaTeX rev 4917), ... produces {{{
\y=macro:
->2.0pt minus 1.0pt. }}} which should be 2.0pt minus 1.0fil, even the above patch is applied.
That is fixed in LuaTeX rev 4952.
Fixed in Release 20140504.0.
There is a MWE.
The result is> \x=macro: ->2.0pt plus 0.1pt minus 1.0pt. l.8 \show\x ? > \y=macro: ->2.0pt plus 1.0pt. l.12 \show\yThe \x is fine. But I think tha the value of \y should be 2.0pt minus 1.0pt to correspond to \ltjsetparameter.It seems that the shrink value of kanjiskip can't be set without a stretch value. I have searched luatexja-en.pdf for detail. But I can't find anything about it. Is this intentional or I miss something?