[XeTeX] Re: [MacTeX] XeTeX

[Jonathan Kew]

On 14 Apr 2004, at 2:41 pm, William F. Adams wrote:

> On Wednesday, April 14, 2004, at 08:24  AM, Jonathan Kew wrote:
>
>> And I'll see if I can do anything about the bloated PDFs that are 
>> generated when you use CM fonts, as someone reported. To some extent, 
>> I think this is an Apple issue, as the xdv2pdf tool generates PDF by 
>> simply imaging with the Quartz engine to a PDF destination. But I may 
>> be able to improve how it behaves.
>
> I was thinking of suggesting a quick-dirty fix, but I think I would 
> probably embarrass myself. I'd be glad to test anything you manage.
>
> Has anyone looked into work-arounds like dumping to PostScript and 
> re-distilling?
>
> Have you considered an option for not embedding the fonts? Then you 
> could print to .ps from Preview or TeXshop and distill from the .ps 
> --- Acrobat Distiller will certainly be able to minimize font 
> inclusion and subset appropriately then.

I don't know what would happen if you print to PS from Preview or 
TeXShop, and then re-distill; perhaps Distiller would be able to 
optimize things. But the original, huge, PDF would still be painfully 
slow to generate and view, so that's not really a useful solution.

I have no real control over font inclusion; this is PDF as generated by 
the Quartz imaging system targeting a PDF destination, and it seems 
that the CM fonts are being embedded separately for each string that's 
rendered using them. :-( I may be able to change the APIs I'm using in 
xdv2pdf, though, and get different behavior. Right now, I'm using 
CoreGraphics APIs to draw the OTF CM fonts, and ATSUI APIs to draw 
using installed Mac OS X fonts, and they seem to behave differently 
with regard to font inclusion.

>
> Also, Marcel Weiher's pdfcompress tool for Mac OS X might be of 
> service to some people.
>
>> I'm also painfully aware of the need for documentation, installer 
>> improvements, etc.; this is only a beginning. Glad to see there are 
>> people interested in it.
>
> Hey! I've been hawking it for years now!
>
> Really glad it's finally available ;)
>
> (I'm not sounding like Lawson English am I?)
>
> BTW - it works fine on Mac OS X 10.2.8 AFAICT, the non-install I 
> reported initially was an error on my part, not fully cleaning up 
> after a deleted Fink install.
>
> One thing I'd like would be some sort of ``\special'' mechanism to 
> access an arbitrary character slot --- the ``the'' ligature does _not_ 
> get used in Jaguar, and in Panther only in the middle of a line.
>
> Is there something along the lines of a ``\noboundarychar''?
>

That would depend on how the AAT tables in the font are set up. In 
principle, you could use the U+200C ZERO WIDTH NON-JOINER and U+200D 
ZERO WIDTH JOINER characters to affect ligature formation, but I 
suspect few fonts actually recognize ZWJ as requesting more ligation. 
(ZWNJ to inhibit default ligatures is more likely to work.)

You can put any arbitrary Unicode character up to U+FFFF in your 
document using ^^^^hhhh, where 'hhhh' are four (lowercase) hex digits, 
or using \char. Note that characters above U+FFFF are not directly 
supported; it should work to use surrogate pairs to access them (XeTeX 
is purely UTF16-based internally), but you can't specify things like 
\lccode for a supplementary-plane Unicode value.

However, if you were really asking for a mechanism to access an 
arbitrary *glyph* from a font, not an arbitrary *character*, that's 
different; no, there's no such mechanism at this point.

Jonathan

_______________________________________________
MacTeX mailing list
http://tug.org/mailman/listinfo/mactex