[OS X TeX] The Font Cache Problem

[Herbert Schulz]

On Apr 13, 2008, at 11:05 AM, Richard Koch wrote:
> Folks,
>
> Many of you are aware of the following bug: you'll be running one or  
> more TeX front ends and/or utilities and suddenly the mathematical  
> symbols will disappear in the preview output. Further inspection  
> shows that a different font without these symbols has been  
> substituted for the correct font. Once one application has the  
> problem, other TeX apps also have font problems, even if closed and  
> restarted. To fix the problem, it is necessary to reboot the machine  
> or use a utility to clear the font cache.
>
> The problem seems to be caused by incorrect data written into the  
> font cache by Apple system software. Most users never run into the  
> problem, while others have persistent trouble. We have inspected  
> source files from the unfortunate users, and the sources appear to  
> be absolutely straightforward TeX code using standard TeX fonts.
>
> This morning a user provided a possible clue. I'd like people on  
> this mailing list (with the problem) to see if his clue is  
> consistent with their own use of TeX.
>
> This user had two machines running Leopard, a PPC and an Intel  
> machine. He only had the problem on the Intel machine. If he  
> rebooted and just ran a few standard TeX apps, he was fine. But he  
> used Equation Editor and/or Equation Service for a while, the  
> problem would surface. (Is Equation Service a separate app, or just  
> a service provided by Equation Editor?)
>
> Then he noticed that his Equation Editor used PPC code and was  
> running under Rosetta. He managed to find the source code and  
> compile it under Intel, and then the problem vanished.
>
> He concluded that the real culprit is Rosetta. I tend to agree.  
> Cocoa apps call Apple system software to image PDF data, and that  
> system software in turn calls font machinery. So it is very hard to  
> imagine how a bug in application code could cause problems in the  
> font cache. But a Rosetta bug makes perfect sense.
>
> If your own experiences confirm this connection, we'll finally be  
> able to report the problem to Apple.
>
> Dick Koch
> koch at math.uoregon.edu
>

Howdy,

I assume the problem with the latest version of LaTeXiT (1.14.4?)  
running under Leopard has nothing to do with this since it's already a  
Universal application. It would be nice if Pierre would fix that so  
Leopard users don't have to run the previous version.

Good Luck,

Herb Schulz
(herbs at wideopenwest.com)