# Re: interesting ideas --- interesting ideas

• To: math-font-discuss@cogs.susx.ac.uk
• Subject: Re: interesting ideas --- interesting ideas
• From: alanje@cogs.susx.ac.uk (Alan Jeffrey)
• Date: Fri, 6 Aug 93 14:03 BST

```>OK, good point. How about two fontdimens x and y, such that the ratio
>x/y gives the needed character number.

That's just asking for rounding errors when the font is loaded, since
the `obvious' code:

\tempcounta=\fontdimenX
\tempcountb=\fontdimenY
\divide\tempcounta by \tempcountb

may end up being off by one.  If you're going to use font dimensions
to record discrete data, just about the only safe way to do it is to
record one of three values: negative, zero or positive!

>Interesting also, that TeX does
>not scale \fontdimen 1. Thus another approach for math fonts might be
>to use fontdimen1 to give the value of \skewchar, and use the ratio
>height/width of skewchar to give the slant per point.

Eeek!  This would need to be over-ridden if the font was a T1-encoded
multi-letter identifier font.

>But actually, it occurs to me now, a better approach is to mandate
>always \skewchar = 0 for math fonts and leave font position 0 empty.

I'd go for this.  Glyph zero is `acute' in T1, and I doubt many Cork
fonts include kerning with acute, so we're probably OK.

Alan.

```