[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: defining the script sizes via font dimens



Frank Mittelbach wrote:

>    > ... but the calculation  *does* establish a linear relation between the 
>    > three sizes. 
>
>   correct. for that reason it is only a fallback.
>
>    > The size don't have to scale linear with the new system,
>    > since the appropriate sizes can be specified separately for each 
>    > body size (the scaling of font dimensions is only relevant if you
>    > load a font 'at' a different size).
>
>   true, but only for fonts that have different design sizes right? so
>   about how many font families we are talking about?
>

 Thats true. I didn't think about that. 

[...]

>   I see the argument that this is information that should ideally come
>   with the font. at what point is the new interface acting? (sorry i
>   haven't yet downloaded your new distribution, i can't do this form
>   this account --- but will next week).right now i don't see a good way
>   how this should work all together, perhaps you have found one.

I have rewritten the NFSS macro \calculate@math@sizes to first load the 
mathfonts only in text size, then get the new font dimensions, set the 
math spacing and script and scriptscript sizes from them (this includes
converting from pt to mu) and finally load the math fonts in all three
sizes. Since \calculate@math@sizes will only be called if the sizes are not
already known, you can still use \DeclareMathSizes to get the sizes you want.
My code seems to work well now, since I am rounding font dimensions to
three digits now, thus avoiding the rounding problems between vptovf and
TeX. 

>   well, first of all i didn't want to say that it is useless, my comment
>   about worth the effort was because you seemed to run into problems on
>   that front. so my pragmatic approach would have been to code default
>   \DeclareMathSizes into the package setting up the new math font
>   interface, for example, by option files, eg
>
>   \usepackage[lucida]{newmath}   % internally loads nmlucida.cfg or something
>

Yes, an interface like that will be unavoidable for families which come
with just one design size. 

>   second, YAASP is not bugfree, you discovered some along the way, it is
>   a student work who did a great job with a lot of help from experts but
>   nobody is perfect. and after all this work was done some time ago and
>   we have learned new facts, for example, that adding parameters to
>   ecfonts wasn't yet a great success (the situation in the math font
>   case is slightly different as we can assure that *all* math fonts
>   usable for TeX will have those additional parameters if used under the
>   new encodings)

A question out of ignorance: what new font dimensions have been added to EC?

>   third, why assume that i might not be wrong? 

Oh, I would never :-)

PS.: And what about the new font dimensions for math spacing (ie \thinmuskip
and friends) ? Has anybody experiences with customizing those values ? Should
we expect them to vary between font families ?