[Fontinst] Re: [tex-fonts] Re: [texhax] On the proper look of the \AA

[Vladimir Volovich]

Hi!

 >> We are discussing the case when there is NO Aring glyph in the
 >> LaTeX font encoding, such as OT1 or T2A font encoding.
 >> 
 >> In this case, we HAVE TO construct the Aring glyph "artificially",
 >> and we have two ways of doing that: either just apply the ring to
 >> A using an \accent command, or put the ring closer to A using the
 >> approach currently used in ot1enc.def.

 > You have ignored one detail: for these fonts the designer can
 > construct the ring and other accents and PLACE THEM AT THE PROPER
 > HEIGHT in order to have their appearance in all composed glyphs as
 > best as possible.  BTW, it's a common principle for font designers.

you mean that e.g. to achieve the "gapless Aring" using the \accent
command, it would be possible to put the ring accent somewhat lower?
are you sure that this approach will work properly for both small and
capital letters?

(If it would be possible to do that, i wonder then why Don Knuth
didn't use this approach when he wanted to create a special look for
the Aring glyph. And why did he decide to make Aring a special case in
the first place)

 >>> BTW, I tried to build the \AA glyph for four fonts from
 >>> AntiquaPSCyr family using the strict standard LaTeX algorithms
 >>> (with the special definition for \AA glyph), and found an ugly
 >>> appearance in some cases because I can see an upper part of the A
 >>> letter inside the ring.
 >> 
 >> May it be the result of incorrect metric for A and/or ring glyphs?

 > No, the metric of all glyphs was corrected.

for which cases did you see such effect? is it reproducible in current
beta version of pscyr?

 >>> So, the special definition is good for CM fonts and maybe for
 >>> some other fonts,
 >> 
 >> Do you have a complete list of "other fonts"? Wouldn't it look
 >> ugly if we were dependant on font families in the generic encoding
 >> definition files?

 > I didn't check all the fonts from the PSCyr collection yet.  I'll
 > do it in a few days.

No, i meant that it's impossible to gather the complete list of font
families to which such trick with composite command should be applied;
and the encoding definition file is definitely not the proper place to
put all such low-level stuff which deals with font families.

I.e. either this trick shall be used for all families (including CM)
or not used at all.

 >>> but is definitely not good for an arbitrary font.  That's why I
 >>> would propose to restrict the application of the special
 >>> definition of \AA only to CM and CM-related fonts.
 >> 
 >> I see no way to do that properly.

 > The ugly position of the ring in composed Aring was surprising to
 > me too.

again, i mean that i see no way to properly do what you have
suggested: i.e. to restrict the application of the special definition
of \r{A} only to CM and CM-related fonts.

it can be left there more or less like it is now, or it shall be
removed. then, \r{A} in T2A encoded CM fonts (LH fonts) will look
differently from \r{A} in OT1 encoded CM fonts.

 > I think this can be the result of different A letter and the ring
 > sizes in different fonts (bolder fonts need larger accents).  BTW,
 > the A letter in AntiquaPSCyr is about 5% higher then the CapHeight
 > and it cannot be lower because of the font style.  In bold variant
 > of the font the ring in composed Aring looks better because the A
 > letter in this font is lower.  I guess the special algorithm under
 > discussion uses the CapHeight instead of the proper letter height.

the definition is somewhat more complex - see the definition. (and
recently, it has been corrected in plain TeX and in LaTeX to use "!"
instead of "h").

ok - as i've seen some comments from Lars Hellström where he says that
he wouldn't have used this trick as "it introduces some rather odd
extra conditions on the font, without giving anything in return for
the indended target audience", i'm inclined to agree to remove it
completely (even for cmr family of LH fonts)...  after all, this IS a
hack, and to get the proper appearance of \r{A} it is better not to
use the font encoding which doesn't contain this glyph.

Best,
v.