Re: Dense encoding, part II

[Mojca Miklavec <mojca.miklavec.lists@gmail.com>]

On 2/10/06, Adam Lindsay wrote:
> Mojca Miklavec wrote:
> > 1. Would Caron & similar uppercase accents make sense? I doubt that
> > many accents are needed in addition to what is already present in the
> > other encoding anyway, but something like that could be used if there
> > is no Ccaron present in the font for example:
> >
> > \definecharacter Ccaron {\buildtextaccent\textCaron C}
> > instead of
> > \definecharacter Ccaron {\buildtextaccent\textcaron C}
> >
> > In well-designed fonts (including all Polish fonts such as lm,
> > antykwa, iwona, ...) the lowercase and the uppercase variant of the
> > accent differ. (Try to write \Scaron\Ccaron in texnansi encoding for
> > example to see the difference).
>
> Good point... except that there are *no* accents available in eurolett,
> anyway. It *should* have all of the accented uppercase characters you
> need (within roman ;). The whole theory is to do away with building text
> accents. But what does Hans want? Should lc and uc accents be available
> to create `weird' combinations?

I know that there are no accents available in "el", but I thought this
was in order to make more space for other glyphs that require kerning
and hyphenation. I still think that people might need to build weird
accents every now and then. At least I would need to - we sometimes
need accents for stressing pronunciation (most of them are present in
"el" just because other languages consider it to be a separate letter,
but see the very strange definition at the end of enco-def.tex ;) and
it would be a pity to take that strength of TeX away. I prefer having
a couple of badly kerned words (most are OK anyway) than having to
make my own glyphs and asking other people to compile those documents
on their computers.

I don't claim that there's a need to have both uc and lc variants, but
at least the lc variant should be present.

And another point: Antykwa Poltawskiego has it's own encoding defined
for example. I guess they're going to complete the font soon in the
same way as they did with Torunska, but currently ConTeXt uses
modified font, so that the font fits into the ec and texnansi encoding
with some empty slots in it. There is no "Ccaron" present in the font,
but I could imagine that ConTeXt could use ANTP. with "el" encoding
and once it would notice that there is no Ccaron present, it would
build it from C and Caron.
I'm not mentioning that because of ANTP since it will be improved
anyway, but because of other incomplete fonts that might profit from
the presence of such accents.

> > Of course some care has to be taken, so that it will also work for
> > fonts without those additional accents for uppercase characters (using
> > \iffontchar perhaps?).
>
> Indeed. I do want to avoid a strong dependency on the specific glyphs
> that appear in the font. That moves the encoding mess to *within*
> ConTeXt, which is not pretty, either.

It depends if it's automatic or not. If ConTeXt could automatically
recognize which glyphs are (not) present, it would be possible to
"play safer".

> I don't hear a lot of calls for
> them since ConTeXt moved to EC as a default encoding.

People who need them stil use texnansi ;)

But sure, I just wanted to point out some glyphs that some might find
useful, I don't claim for any of them that they should go in (except
for accents and ordfeminine/masculine/numero), they might need
discussing first.

If we're talking about Adobe: many fonts include "mu" (micro) or "Ohm"
for typesetting units. In lm and other TeX fonts this is not necessary
since there is math for it. I'm hardly ever using Adobe fonts, so I
can't really judge if this might be useful or not.

Mojca