[Fontinst] ugly Polish diacritics

LarsHellström Lars.Hellstrom at math.umu.se
Tue Feb 25 23:24:40 CET 2003


At 21.22 +0100 2003-02-25, Walter Schmidt wrote:
>On Mon, 24 Feb 2003 23:40:27 +0100, Lars Hellström wrote:
>
>> [re zdot]
>> Using the height of i as a comparison is somewhat
>>risky though, as it will fail for smallcaps. Perhaps the height of
>>odieresis is better?
>
>All I can say, is, that a Polish expert told me that \.{z} must
>match the i.  And there is no odieresis in the Polish language.
>In a smallcaps font, however, we can stay with the present code
>for the Zdotsmall glyph:  Ismall has no dot, so there is nothing
>to "match".  That's how I understand it.

OK, makes sense.

>See also
><http://studweb.euv-frankfurt-o.de/twardoch/f/en/typo/ogonek/>
>a comprehensive overview on the design of the Polish letters.

One point made there seems to be that very many designers get this wrong,
so maybe it isn't only fontinst that should be blamed ...

>Btw, are you going to adopt my zdot for the fontinst distrib?

I was generally hoping that we (the subscribers of this list) could spend
some time and collect improvements on the traditional code of latin.mtx, so
that it can be part of the proper fontinst distribution. A zdot improvement
would definitely fall in this category. I had also intended to include code
some for commaaccent and commaaccented letters, but I have been unable to
find it (I'm pretty sure someone once sent me fontinst code for faking
this, though).

However, I also realise that I do not have the overview of available fonts
that is needed to discover these kinds of problems with the recipes
fontinst has traditionally used. I could certainly need someone (like you,
perhaps) to assist with maintenance in this area.

>
>>[...]
>>>With Eogonek and Aogonek, the ogonek often protrudes over
>>>the right edge of the letter.  This can be avoided by
>>>building the glyphs in a way so that the the right edge of
>>>the ogonek is aligned with the right edge of the letter.
>>
>>This is BTW the default LaTeX definition of \k.
>
>Why not adopt this for the fontinst distrib, too?
>It might be a bit better than the present code.

Indeed, your reference seems to suggest that it should be something like this.

>>>[...] it does not work
>>>with monospaced fonts, where the bare ogonek acent has the
>>>same width as all other characters.
>>
>>Special cases for monospaced fonts are common in fontinst.
>>\ifisint{monowidth}\then ... \Else ... \Fi constructions are often
>>necessary.
>
>I know, of course.  I just wanted to point out that there is
>no automatic solution for the monospaced case.

This is possibly why Alan didn't use the "right edge" definition in latin.mtx.

>Generally, even with the suggested fixes, one still cannot
>create usable Polish letters without manual interaction.
>For instance, I have made the font metrics for the LucidaBright,
>LuxiMono and URW Grotesk fonts on CTAN.  Having to check and
>correct all the aogonek's and eogonek's -- and in the typewriter
>fonts also the corresponding uppercase letters -- manually,
>that's not an amusing idea!  I hope you don't expect me do do
>that...  Any volunteers (perhaps from Poland)?

One would think they ought to be the ones most interested in this.

>
>>[...]
>>>For instance, my VFs for the DayRoman font
>>>were created with fixes such as
>>>
>>>  \resetglyph{aogonek}
>>>     \botaccent{a}{ogonek}{412}% originally {850}
>>
>>I thought the ogonek was supposed to hang from the stem/serif of a
>>rather than from the bowl?
>
>Yes, but I cannot explain why a displacement of 412 was appropriate
>in this case.  Take a look at <http://home.vr-web.de/was/x/dayrom.pdf>
>and judge it yourself!

It certainly looks right. Could this be a case of an "ogonekfitted", i.e.,
the width of the ogonek is adjusted to the width of some letter (cf.
ringfitted), and is quite unlike the width of its bounding box?

Lars Hellström




More information about the fontinst mailing list