[XeTeX] some odds-and-ends questions

[Will Robertson]

On 19 Jan 2005, at 8:00 PM, Jonathan Kew wrote:

> Color as part of the font definition seems a natural and useful 
> approach, to me, but I realize it doesn't integrate as easily with 
> existing packages.

But the advantages are outlined below as you say: the colour is 
inherent in the glyph, so bugs in eTeX or where-ever won't mess things 
up, ever. I consider this A Good Thing™.

> <informative piece about how colour works; thanks!>
>
>> <about the XeTeX logo>
>
> Yes, I've done a similar thing in a document recently, to prevent the 
> logo disturbing the leading. Otherwise, if your font size is close to 
> the baselineskip, it's likely that with the raised and lowered boxes 
> (and XeTeX's version of their height and depth, based on overall font 
> metrics rather than true glyph sizes), you'll get lineskip kicking in.

I realised that was why you put it in the zero height/depth box (I 
remember a discussion on such things a while ago), but I wasn't sure 
what it was about the TeX commands you were using that introduced the 
extra space. (Just out of curiosity...)

>> Finally, I recall that XeTeX has no real knowledge of the height and 
>> depth of a glyph. For example, \settoheight\xheight{...} of any 
>> character in a font will give an incorrect number. Also, 1ex 
>> (\fontdimen5) seems to be equal to 3.9998pt no matter which font is 
>> chosen (in a 10pt document).
>
> Right. Currently, XeTeX just stuffs some rather arbitrary values into 
> the \fontdimens based on the ascent and descent of the font as 
> reported by ATSUI, and the width of the space character.

I've noticed a couple of fonts seem to have rather large interword 
skip, but I might have neglected to adjust the \linespread to even 
things out. I didn't investigate very much, primarily because I'm a 
very very amateur typesetter!

>> <adding real \fontdimen5 info>
>
> It's very reasonable to want this sort of thing, and I hope to make 
> some progress towards it. But what would we do about setting 
> \fontdimen5 if the font doesn't even have an 'x'? Imagine a 
> Devanagari-only font, for instance.

Ha ha ha! Right you are. Um, that one's tricky. Well, I might suggest 
you couldn't do worse than what is done at the moment. If a font *does* 
have an "x", it can have a proper x-height; if not, give it the default 
value like you do now? I would assume in Devanagari typography, an "ex" 
unit wouldn't ever be used anyway, but I wouldn't even be able to 
recognise Devanagari from Tibetan (well, I could probably guess) so I 
might be very misguided.

Regards!
Will