Ann: fontinst v1.9 prerelease

[Lars Hellström]

Hilmar Schlegel wrote:
>What do you think about composites using a character bbox which is
>enlarged like in the vertical direction also in the horizontal direction
>if the accent lies outside the base bbox? If this behaviour is used only
>for those glyph-not composites which must be composed in the VF this
>would be of course against the convention how Type 1 composites actually
>work in a printer but it allows to specify dquoteright and such as
>normal composites in AFM files. I use this feature without interference
>with other common composites since in most cases the accent is inside
>the base character's width. The geometrically placed x-quoteright is
>usually less acceptable.

Frankly, I don't see what you mean. Width, height, depth, and italic
correction of glyphs generated from CC instructions in an AFM file are all
set to exactly the values that the corresponding "-not" glyph have; none of
them take any notice of the internal structure of the glyph.

>What about a \setkern command which respects the minimumkern (sometimes
>smaller kerns might be usable for positioning by fontinst but in general
>the smaller kerns are dropped anyway)?

I have some problem seeing why \setkern should look at \int{minimumkern}.
You might save some memory by doing it, but I don't think it would be that
much since small kerns have already been removed once in the AFM to MTX
conversion. Still, it you absolutely want one, you can always use the
following definition:

  \setcommand\mysetkern#1#2#3{%
     \ifnumber{\int{minimumkern}}<{\abs{#3}}\then
        \setkern{#1}{#2}{#3}%
     \fi
  }

>If standard stems are set in the fontdimens I suggest to use the StdVW
>from AFM 4.0 files instead of the width of I.

It does sound like the right thing to do to me, but I'm not that acquainted
with the format of AFM files or the fontdimens used. It seems proper that
Thierry Bouche should give a comment on this (since he seems to have made
the last major update of fontinst in this area).

>
>>   * AFMs with non-integer dimensions can now be read correctly.
>
>The top feature would be to be able to read character bboxes and do
>positioning not only based on the advance width. There would be of
>course no restriction if Tex then gets only the width-based metric for
>doing its part of typesetting - just positioning in VFs like for
>monospaced fonts would be more flexible.

There is a feature in v1.9 which gives you some access to the bounding box,
check out the description of \italcorr_expression.

Lars Hellström