Re: fontinst 1.801 -- still worse

["Berthold K.P. Horn" <bkph@ai.mit.edu>]

At 01:27 PM 6/9/99 +0000, Sebastian Rahtz wrote:

>  a) most people don't have a menu with "WriteTFM" on it. It seems
>reasonable, in the TeX world, to promote portable, free,
>solutions. People who want to can buy better solutions from you, but I 
>want to cater for the lowest common denominator

Well, my idea is that AFM2TFM be revived to where you can use it
to make TFM files complete with ligatures and kerning pairs.
As you may have noticed, I have been lobbying for a lot of things
that should happen that do *not* involve buying anything.  For
example, having all TeX implementations read all three flavours
of line termination.  Don't need to buy Y&Y TeX for that :-)!

>  b) in our `free' world, dvips cannot grok TrueType fonts; thats
>probably a far bigger barrier to common use than the metrics

How is this relevant?  I thought we were talking about Type 1 fonts.
WriteTFM is not restricted to TrueType fonts -- it works for
any font format properly installed in Windows (which these days
is just three: Type 1, TrueType and OpenType).

>  c) nobody demands that you use fontinst. afm2tfm is simple and easy
>to use, like WriteTFM. fontinst is there if you want its extra facilities.

Small changes to AFM2TFM would make it actually useable.
Namely making it able to do kerning and ligatures.

>  > Why cripple AFM2TFM so it can't be used for "normal" use?
>  > (Fortunately there is a work around: 
>  > http://www.yandy.com/maketfm.htm).

>interesting. i wish afm2tfm had more control though, over things like
>the size of spaces.

>  > I use AFMtoTFM and gain free time that way to do *real* fun things :-)
>  
>does WriteTFM just call "AFMtoTFM"? or does it have its own converter?

Both.  It first has to create a suitable AFM file (with added comments)
and then call AFMtoTFM.  How it links to AFMtoTFM (DLL or EXE) is less 
significant.  To create a good AFM file it has to trace out all the glyph 
outlines, follow calls to construct composites from base characters etc. 
and do this for the different font formats.

>incidentally, i think we should all agree to set checkums to 0 in all
>our metrics, to avoid these questions about them...

I disagree, I think we should (i) turn off meaningless complaints
from current versions of applications and (ii) use the radix 40 scheme
described elsewhere to hide the name of the encoding in the checksum
for advanced applications that can check this :-)

Regards, Berthold.

Berthold K.P. Horn	mailto:bkph@ai.mit.edu  http://www.ai.mit.edu/people/bkph/bkph.html