Why is \AA still defined as it was in LaTeX2.09?

Question

The \AA and \aa macros are defined in latex.ltx as

\def \aa {\r a}
\def \AA {\r A}

This fails with the X2 encoding which has no Latin letters. Why didn't the LaTeX team revert to the following definition:

\DeclareTextSymbolDefault{\AA}{OT1}
\DeclareTextSymbolDefault{\aa}{OT1}
\DeclareTextCommand{\aa}{OT1}{\r a}
\DeclareTextCommand{\AA}{OT1}{\r A}

Here is a MWE (UPDATED):

\documentclass{article}
\usepackage[utf8]{inputenc} % source file encoding
\usepackage[X2,T2A]{fontenc} % cyrillic font encoding
%  \DeclareTextSymbolDefault{\AA}{OT1}
%  \DeclareTextSymbolDefault{\aa}{OT1}
%  \DeclareTextCommand{\aa}{OT1}{\r a}
%  \DeclareTextCommand{\AA}{OT1}{\r A}

\begin{document}%

\verb|\AA|=\AA
\verb|\aa|=\aa

\fontencoding{X2}\selectfont

\verb|\AA|=\AA
\verb|\aa|=\aa

\end{document}

Good question, but for mixed input like that I'd just use a unicode engine and a font which supports both scripts and input Å directly. — Henri Menke, Aug 08 '17 at 03:51
X2 is not a “proper” LaTeX font encoding, for the precise reason it doesn't contain the Latin alphabet. — egreg, Aug 08 '17 at 08:28
as egreg says the reason it is X.. not T.. is because it doesn't follow the rules that underlie the assumptions in the code with respect to latin in the ascii positions and T1-compatible upper/lower case tables. That said something like you suggest would be possible but I have to ask, why would you expect \AA to work, if A does not work within the scope of that encoding? — David Carlisle, Aug 08 '17 at 09:39
you do not show your output but I get this with tl2017 which seems to show \AA working OK??? — David Carlisle, Aug 08 '17 at 09:44
@egreg: X2 encoding contains ancient slavonic letters. There are no enough slots for all such letters in T2* encodings. — Igor Kotelnikov, Aug 08 '17 at 11:30
@DavidCarlisle: I have exactly same output as in your answer. — Igor Kotelnikov, Aug 08 '17 at 11:31
@IgorKotelnikov so the example posted does not show any problem, you could have used an example that did not load babel-russian it woudl have been rather clearer. — David Carlisle, Aug 08 '17 at 11:32
@DavidCarlisle: @DavidCarlisle: I have exactly same output as in your answer is case of russian-babel available from CTAN. But I experimented with forthcoming version where workaround for X2 encoding is removed. I believe that russian-babel is not a good place for such workaround. That's where my question raised from. — Igor Kotelnikov, Aug 08 '17 at 11:42

David Carlisle · Answer 1 · 2017-08-08T12:08:02.153

In LaTeX 2.09 the definitions were

\def\aa{\accent23a}

\def\AA{\leavevmode\setbox0\hbox{h}\dimen@\ht0\advance\dimen@-1ex%
  \rlap{\raise.67\dimen@\hbox{\char'27}}A}

as there was no encoding support, and no \r accent defined.

So whatever else is the case \aa and \AA are not defined as in LaTeX 2.09.

With current latex and your MWE shows that the effective definition in your case is

> \AA=macro:
->\OT1-cmd \AA \OT1\AA .

and you do get an accented A in X1:

\documentclass{article}
\usepackage[utf8]{inputenc} % source file encoding
\usepackage[X2,T2A]{fontenc} % cyrillic font encoding
\usepackage[english,russian]{babel}
%  \DeclareTextSymbolDefault{\AA}{OT1}
%  \DeclareTextSymbolDefault{\aa}{OT1}
%  \DeclareTextCommand{\aa}{OT1}{\r a}
%  \DeclareTextCommand{\AA}{OT1}{\r A}


\begin{document}%

\verb|\cyrillicencoding|=\cyrillicencoding

Русский English

\meaning\AA

\verb|\AA|=\AA
\verb|\aa|=\aa

\section{X2}

\fontencoding{X2}\selectfont
\verb|\cyrillicencoding|=\cyrillicencoding

Русский English

\show\AA

\verb|\AA|=\AA (A \r A)
\verb|\aa|=\aa (a \r a)

\show\AA
\end{document}

The suggested definitions would need to be extended to all encodings.

As posted (and as currently in Russian babel) they break \AA for (for example) T1 encoding which is bad as it is the main encoding for the Scandinavian languages for which this is used.

\documentclass{article}

\usepackage[X2,T1]{fontenc} % cyrillic font encoding

\showoutput

\setbox0\hbox{foob\aa r}\showbox0

  \DeclareTextSymbolDefault{\AA}{OT1}
  \DeclareTextSymbolDefault{\aa}{OT1}
  \DeclareTextCommand{\aa}{OT1}{\r a}
  \DeclareTextCommand{\AA}{OT1}{\r A}

\setbox0\hbox{foob\aa r}\showbox0

\begin{document}%





\end{document}

Shows that by default \aa in T1 comes from the T1 encoded font, which allows hyphenation and kerning to work correctly

\hbox(6.8872+0.0)x27.79877
.\T1/cmr/m/n/10 f
.\T1/cmr/m/n/10 o
.\kern0.27771
.\T1/cmr/m/n/10 o
.\T1/cmr/m/n/10 b
.\T1/cmr/m/n/10 �
.\T1/cmr/m/n/10 r

But after the suggested definitions \aa comes from the OT1 font

> \box0=
\hbox(6.94444+0.0)x27.8
.\T1/cmr/m/n/10 f
.\T1/cmr/m/n/10 o
.\kern0.27771
.\T1/cmr/m/n/10 o
.\T1/cmr/m/n/10 b
.\kern -1.25 (for accent)
.\OT1/cmr/m/n/10 ^^W
.\kern -6.25002 (for accent)
.\OT1/cmr/m/n/10 a
.\T1/cmr/m/n/10 r

So hyphenation in any word using å is broken.

@IgorKotelnikov I have updated my answer with an example showing problems with the suggested definitions — David Carlisle, Aug 08 '17 at 12:08

score 6 · Answer 2 · answered Aug 08 '17 at 10:37

6

If you look at russianb.ldf, you can see

 86 \expandafter\ifx\csname T@X2\endcsname\relax\else
 87   \DeclareTextSymbolDefault{\AA}{OT1}
 88   \DeclareTextSymbolDefault{\aa}{OT1}
 89   \DeclareTextCommand{\aa}{OT1}{\r a}
 90   \DeclareTextCommand{\AA}{OT1}{\r A}
 91 \fi

(line numbers added for reference), which means that, provided the X2 font encoding has been initialized in the document, the \AA and \aa commands will do what you propose. Otherwise there is no need for this, as an encoding whose name starts with T is supposed to have Latin letters in the corresponding ASCII positions.

answered Aug 08 '17 at 10:37

egreg

1,121,712

I am maintainer of babel-russian and my question was initiated by my desire to remove the indicated lines from russianb.ldf. I believe that russianb.ldf is not a proper place for such sort of definitions. They should be moved, eg, to x2enc.def. – Igor Kotelnikov Aug 08 '17 at 11:23
2

@IgorKotelnikov you could have mentioned in your question that the MWE you had posted did not show the problem:-) The problem with the definitions is that it leaves \AA wrong for other encodings, eg T1, would need a similar definition or \AA would come from OT1 – David Carlisle Aug 08 '17 at 11:31
2

@IgorKotelnikov You should mention why you'd need \AA when the X2 encoding is active. – egreg Aug 08 '17 at 12:21
@egreg: This is 2nd reason why Russian babel is not a proper place for definition of \AA: X2 encoding lacks all latin letters so why there should be an exclusion for \AA? So I almost for sure will remove this chunck of code from next release of Russian babel. – Igor Kotelnikov Aug 08 '17 at 13:18

Why is \AA still defined as it was in LaTeX2.09?

2 Answers2