[tex-live] glyphlist.txt

Eddie Kohler kohler at cs.ucla.edu
Sat Jun 11 18:57:48 CEST 2005


Hi Karl,

Karl Berry wrote:
>     I don't know what the TeXLive distribution does instead.  
> 
> Nothing.  So the file is not found, unless it happens to exist in the
> user's /usr/local/share.

Ehwhoops.

>     Probably I should update to search using kpathsea when available.
> 
> Yes, I think so.  The simplest thing would be to do the same thing that
> dvipdfmx does, and search using "other text files".  That will end up
> looking in $TEXMF/$progname// for it, I believe.  

OK.  The code will (1) look under "other text files", then if not found, (2) 
look in PREFIX/share/lcdf-typetools ($pkgdatadir).

The makefiles will not, however, install glyphlist.txt in a TDS.  You'll have to 
do that by hand.

> Which leads to more questions:
> 
> 1) Are there other data files you search for now?

No.  lcdf-typetools installs 7t.enc in PREFIX/share/lcdf-typetools, but just as 
information.

> 2) Do you explicitly set $progname?  If not, it is essentially the
>    basename of $0, so it'll be otftotfm or whatever program is looking
>    for it.  In which case it seems like it would be desirable to set it,
>    which I believe basically amounts to
> kpse_reset_program_name("lcdf-typetools");

OK.  I had been setting it explicitly to argv[0]; will anything go wrong for 
existing installations if I just switch to lcdf-typetools?  Also, lcdf-typetools 
contains a dash, which leads to kpathsea lookup on environment variables that 
contain dashes.  That'll surely cause a problem for someone.  I suggest 
"lcdftools" instead, which is what otftotfm already uses for vendor strings.

> 3) Since I gather you extend the file (with additional code points), we
>    shouldn't try to share it with other glyphlist.txt's even if we
>    wanted to (which we don't).  Right?

That's right, I wouldn't.  The lcdf glyphlist is different from Adobe's in 
several ways.

One of them is actually worth some thought.  In Adobe glyphlist, the name "phi" 
maps to a "loopy" phi, while "phi1" maps to a "straight" phi.  In TeX practice, 
as codified in the encoding files, this is reversed: "phi" is straight, "phi1" 
is loopy.  My glyphlist.txt recodes to match the TeX practice, but this feels a 
little iffy.

I will release 2.34 with these changes if you approve.

Eddie



> Thanks,
> Karl



More information about the tex-live mailing list