# [XeTeX] Ideal way to handle fonts

Jonathan Kew jonathan_kew at sil.org
Mon Jun 26 14:09:30 CEST 2006

On 26 Jun 2006, at 12:50 pm, William Adams wrote:

> Okay, I've been doing some further testing and have some questions,
> and then some more questions.
>
>   - how does the version of w32tex that's available on that web page
> match what's available in SVN?
>   - could we get a marker or build number on that page or something
> to let us know when it's been up-dated?

Akira, would you be able to add the svn revision number to the web
page and/or zip file name when you post a new binary? That would help
people know how the W32 build compares to what they have on other
platforms, and to the svn log entries to see what fixes are supposed
to be present.

Obviously, the version number of xetex itself will change to 0.995
(etc) at some point, but I don't update this for every fix that's
checked in to Subversion; that would be a nuisance during
development, as every time that version number gets changed, the pool
file changes and every format file must be rebuilt, which is a hassle
when experimenting with things that don't actually relate to that
level of TeX.

>   - I got three crashes in w32tex over the weekend, not really
> reproducible (seems to happen when I run successive calls to TeX on a
> file in WinTeXshell (doing this after editing the file to switch
> fonts) --- would someone want the Dr. Watson crash logs and memory
> dumps (the latter is ~14MB, so wants an FTP site to drop it off at)?

I don't know what I'd do with a 14MB dump, but if the log includes a
stack trace showing where it crashed, I'd be curious to see that.

>   - does the w32tex 0.994a binary have the fix to make \setromanfont
> {Aldus LT Std} work? ('cause it didn't).
>
> Several more fonts didn't work correctly (or at all) and others
> exhibited unpredictable behaviour, so how do we want to handle this?
>
>   - Antique Olive for instance set the light as the normal weight and
> the Compact for the bold --- do we want to try to extend or improve
> the handling of weights? Bauhaus performed similarly.

Can you compare the behavior on different platforms, and see if it's
consistent?

> Also, W32tex (or is it XeTeX itself? or the underlying Windows OS?)
> doesn't handle spaces very well. I tried this (where the .tex file
> was named the same as the font I wished to test):
>
> \documentclass{typespec}
> \begin{document}
> \specimenfour{\jobname} % this feeds into a \setromanfont{#1}
> \end{document}
>
> and it complained about not being able to find a file and balked at
> the space in the fontname.

Hmm... probably a xetex problem with quoting the \jobname.... I'll
try to look into that.

JK