On Sun, May 30, 2010 at 04:12:24PM +0200, Manuel Pégourié-Gonnard wrote: > Le 30/05/2010 16:02, Khaled Hosny a écrit : > > I think we should no longer ask people to run mkluatexfontdb > > script, luaotfload will take care of generating database. > > Ok, good to know. > > > Also, AFAIC, the script is the supposed to run from the same directory > > where luaotfload and alt_getopt reside, anything else is not tested. > > > This restriction obviously needs to be removed for production use. I think it > should work now, with the new lua search path in TL (so, with tomorrow's > update > of tlpretest).
It was not intented as a restriction, but it is the only case I test and is known to work, others are free to test other use cases and fix what is broken. > > > No real multiuser support either, there is a --sys option but it does > > not work as advertised (i.e. it will index fonts that only accessible to > > the user running it), and I don't personally support it either. > > > So, maybe it's preferable to remove the option (and the corresponding items in > TL: postaction, tlmgr button...) and update the doc. > > I think, however, that it's worth keeping the mkluatexfontdb executable in the > bin dir, so that the user can run it with options (eg, -v for debugging). > Wdyt? It is up to Elie to decide, I never likes the --sys version, but he seems to have a good reason for it (--destdir=/path/to/systexmfvar should have the same effect, but at least don't give the illusion we are multiuser.) > > That is my fault, I should have run it before uploading to CTAN. > > > I added a "check" target in the Makefile and made "ctan" depend on it (a good > habit for every package, if you want my opinion), so as to avoid future > similar > problems. It is a bit slow, but it is fine, it would makes such mistakes hard. -- Khaled Hosny Arabic localiser and member of Arabeyes.org team Free font developer
