Misplaced code

Hilmar Schlegel Hilmar Schlegel" <schlegel@vossnet.de
Wed, 25 Nov 1998 22:59:02 -0500

Lars Hellström wrote:
> "Hilmar Schlegel" <schlegel@vossnet.de> wrote:
>  [snip]
> >i) it is not plausible that the kerning of word-boundaries of a non-Tex
> >font applies to the kerning of the visiblespace
> Note that the current latin.mtx does make visiblespace kern like space.

I wanted to point out that this is questionable since as far as non-Tex
fonts are concerned, kerns with <space> are intended for what has to be
translated into boundarychar in Tex. Aside from the plausibility to use
visiblespace as boundarychar it is questionable if kerns intended for a
boundarychar work actually also for visiblespace.

> I was simply pointing out one possible reason that the code in question
> existed in the first place, I did not take a side on whether this reason
> was good or not.

I assumed it might be worth to point to this problem while this topic is
hopefully clarified.

>  [snip]
> >The following from t1.etx is quite clumsy due to the explicit code
> >position but works at least:
> >
> >\ifisglyph{compwordmark}\then
> >%   \setint{boundarychar}{\int{visiblespace}}
> >   \setint{boundarychar}{23} % \int{compwordmark}} %124
> >\fi
> >
> Excuse me, but isn't compwordmark meant to be used in words, to split up
> possible ligatures in composite words like ``shelfful''? I do not know if
> anyone actually uses it this way, but I am pretty sure that is how it is
> intended to be used.

Perhaps, only it is a matter of taste to consider ligatures as
determined by the actual geometry of characters or by the meaning of the
words typeset. If really anybody has the desire to do the markup
overhead then one could ask why he is not using Rif\/fischer just in
case. Anyway if cwm has any other practical relevance for someone he can
choose something according his fancy: the point was that the quite
natural solution to set the code won't work.
Possibly one has to introduce an explicit character <boundarychar>
instead of only dealing with its character code (plus a flag to include
boundarychar ligatures & kerns or not).

Hilmar Schlegel