[XeTeX] Should xelatex have its own kernel? (was: "Conflict between xunicode and fontspec?")

[Will Robertson]

Hello,

I'm a bit confused about group hopping here, so I'm posting this both  
here and in the unimath group (which is a somewhat dormant list to  
talk about unicode maths in XeTeX+LaTeX).

On 08/02/2008, at 9:20 AM, Ross Moore wrote:

> On 08/02/2008, at 5:24 AM, Joel C. Salomon wrote:
>
>> On Feb 7, 2008 3:28 AM, François Charette <firmicus at ankabut.net>
>> wrote:
>>> U+2007 -> ? "figure space"
>>
>> A space the size of a numeral (if using lining figures; if the font
>> only has proportional figures I'd guess '8').
>>
>> Although... mightn't the font include the width of these various
>> spaces?  I'd suggest using the values from the font, if such exist.

Without knowing the reasoning behind this glyph, I'll go out on a limb  
and predict that this space character is designed to be used in  
tabular material with fixed-width numbers, when a usual space wouldn't  
necessarily be the right width for the text to align vertically  
between lines.

> For example, if a piece of LaTeX source has been constructed
> by copy/paste from an existing PDF or other file, then these
> characters may well be included.
>
> Should XeTeX treat them as a normal space tokens, then let
> the typesetting context determine what to do ?

I think this is an open question. Many of the space tokens in unicode  
have unambiguous meanings so if you see a thinspace in the source then  
it makes sense to keep it for the output. Especially for text  
processing.

> Indeed, should XeTeX be smart enough to *insert* these
> spacing characters into the output that it creates?
> This is particularly relevant to the typesetting of
> mathematics, and it would indeed be a departure from
> the way TeX currently works.

I think we can safely assume that the TeX community knows more about  
mathematical typesetting than any font designer (with a couple  
exceptions). Therefore I'd say it's reasonable not to use the unicode  
characters for these spaces when PDF is the output mode. Most fonts  
don't even have most of the characters, right?

> Against this is the question of how do "smart" math fonts
> handle the spacing? (e.g. the STIX fonts).
> Aren't they supposed to know the correct amount of space
> to put around operator and relation symbols?
> So adding these characters explicitly would be either
> redundant or just plain wrong.

I don't believe there is any font that smart yet.
Although I don't think it's impossible... But what font designer would  
bother, when you need a typesetting program to do at least some of the  
work for you?

I mean, you could approximate good maths typesetting with plain  
unicode text with all these spaces and so on, but you'd still not be  
able to do superscripts or subscripts, or limits, or integrals that  
change size, or square roots, or ... so what's the point?

Will
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2415 bytes
Desc: not available
Url : http://tug.org/pipermail/xetex/attachments/20080208/aa8bd129/attachment.bin