[Fontinst] letterspacing

[Lars Hellström]

At 00.03 +0200 2005-04-18, Michael Zedler wrote:
>Lars Hellström schrieb:
>> At 23.19 +0200 2005-04-17, Michael Zedler wrote:
>>
>>>Or did
>>>I simply misunderstand letterspacing, smallcapspacing, capspacing?
>>
>>
>> AFAICT from my own comments on these, they should all have the same value
>> (when you're at all using the glyphs in question), so there's not much to
>> misunderstand.
>
>(I forgot about smallcapsextraspace)

That one's different. In newlatin.mtx it is used when _faking_ smallcaps by
shrinking the upper case, but only then. It thus only affects the *small
glyphs. It is also worth observing that smallcapsextraspace is applied
twice to each glyph (the width is increased by 2*smallcapsextraspace),
whereas letterspacing is applied once.

The coding of this was more involved in latin.mtx, but with newlatin.mtx
there shouldn't be a need to coordinate letterspacing and
smallcapsextraspace.

>Setting all to the same value gives
>sometimes both the letterspacing and the other-spacing.
>
>lsfake:
>-   \setint{smallcapsextraspace}{25}
>+   \setint{smallcapsextraspace}{0}
>and delete the definition of
>
>lsmisc:
>-\setint{smallcapsspacing}{0}
>+\setint{smallcapsspacing}{\int{letterspacing}}

That wouldn't work in general. letterspacing is normally unset at that
point. It would have to be conditioned on \ifisint{letterspacing}\then ...

Mind you, there could be an error of policy here. letterspacing has always
been documented as an _encoding_ variable, which is why it isn't used in
the metric files, but that could be the wrong approach. It is certainly a
rather metric quantity, and for glyphs like SSspaced it needs to be fixed
already at the metric stage. That issue thus needs to be reexamined.

>lubuild:
>change capspacing to letterspacing for
>IJspace, SSspaced,
>
>all very confusing, but now the vpl looks ok.
>
>Now I'm looking for a way to set the bounding box of f und j in an
>italic shape to the complete glyph width (like in CMMI). Any idea?

Not sure what you mean here; since fontinst doesn't export any bounding box
data, I don't see much point in setting it. In case you mean it the other
way around, then I'd start by including bbox.sty so that the bounding box
data becomes available. From there on it's a pretty straightforward
fiddling with \resetglyph and friends, but remember that in math some of
the apparent glyph width should be coming from the italic correction.

Lars Hellström