On 20/09/2010, at 2:20 AM, Khaled Hosny wrote: >> I have no idea how xetex is going around. But I found the place >> where it fails: in \...@subst@correction. A failing example gives >> this with tracingmacros =1 : >> >> >> EU2/testfam/m/n -><->"name:Lucida Sans" >> >> <->"name:Lucida Sans" ->
It's very strange that this works with XeTeX, since nothing different is happening internally. > How I forgot that! internally, "Lucida Sans", "LucidaSans", > "lucidasans", "L U ci DaS anS" all are the same to luaotfload, we will > strip spaces and lower case font name before doing matching. I haven't done the lowercasing, but I've just pushed an update to fontspec on Github that zaps all spaces at the very beginning and it seems to fix the problem. (Even if the cause is still somewhat odd.) Would you like to test it out? I introduced another bug in the last official update over the weekend so I need to upload a new CTAN soon-ish. Might take me until the end of the week if things get hectic, though. Thanks for the help! -- Will
