[XeTeX] [OS X TeX] Accented characters in Xe(La)TeX

cfrees at imapmail.org cfrees at imapmail.org
Sun Jun 6 04:23:43 CEST 2010


On Fri 4th Jun, 2010 at 23:18, Peter Dyballa seems to have written:

>
> Am 31.05.2010 um 23:31 schrieb cfrees at imapmail.org:
>
>> Attached: source and pdf
>
>
> The effect is certainly Tiger based.
>
>
> I used the Venturis fonts from TeX Live 2008 and 2009 (the files have not 
> changed between release dates). I cleared font and application caches before 
> switching to the other set. I used the binaries from from TeX Live 2008, 
> 2009, and 2010 pretest. In every case the same result. The result in the 
> created PDF file also shows up now in Leopard and is not different from what 
> you sent, Clea.
>
> So it's likely a bug in Tiger's font service which is triggered by something 
> in the regular font variant.

So it would seem... but it turns out to be more complicated. There *is*
something wrong with the regular versions of the different Venturis
font families. (At least, I only tested with VenturisADF-Regular but it
looks as though it covers the others too.)

Two guesses:
1) Leopard handles fonts differently so the bug doesn't affect it (i.e.
Leopard handles fonts more like Windows, less like traditional Mac);
2) Leopard handles buggy fonts more gracefully.

Here's the problem: the Mac name does not match the Windows name. The
Mac name is "Venturis ADF"; the Windows name is "VenturisADF-Regular".
For bold, things are different: Venturis ADF Bold" versus
"VenturisADF-Bold". Font Book doesn't see the problem but Finder does.
More especially, FontForge does.

I regenerated the otf font using FontForge without making any changes
and got a font which Finder identified as "VenturisADF-Regular".
Disabled the original; enabled my substitute. All weirdness gone.

So the most we can lay at Tiger's door is, I think, that it fails to
handle a certain font bug gracefully. It is not especially clear this
is a bug, though. It might be that Tiger, unlike Leopard, uses the Mac
name routinely. That might give more robust results if there are more
fonts with incorrect Mac names than Windows names, but it isn't
necessarily a bug per se. It does seem to me to be a bug that Font Book
doesn't recognise the problem with the font, however. And if the font's
Mac name matters to the system, why the hell does Font Book only
recognise it's Windows name?!

Cheers,
cfr




More information about the XeTeX mailing list