PCTeX Talk

zedler · Joined: 03 Mar 2006 Posts: 15

Hello,

\documentclass{book}
\usepackage{times,amsmath}
\usepackage[subscriptcorrection]{mtpro2}
\begin{document}
$a_\text{eff}$
\end{document}

The above example gives an error, can this be avoided without putting the subscript into extra braces?

Michael

Michael Spivak · Joined: 10 Oct 2005 Posts: 52

jautschbach · Joined: 17 Mar 2006 Posts: 11

zedler · Joined: 03 Mar 2006 Posts: 15

Michael Spivak · Joined: 10 Oct 2005 Posts: 52

WaS · Joined: 07 Feb 2006 Posts: 27 Location: Erlangen, Germany

I have consulted three LaTeX textbooks for beginners wrt/ this topic:
None of these implies that you may omit the braces after the underscore
(unless there follows one single character only, of course).
Also, I remember having learned that omitting the braces is "bad style",
and that I have to expect that it may not work.

jautschbach · Joined: 17 Mar 2006 Posts: 11

Michael Spivak · Joined: 10 Oct 2005 Posts: 52

OK, here's what's going on.

1. For some reason, amsmath.sty, or more precisely amstext.sty, which
it uses, puts an extra set of braces around the definition of \text. This
has the effect of making a_\text{b} work, even though any one reading the TeXBook, or any LaTeX manual would assume it _shouldn't_ work. I
guess some one thought this was a neat idea because people were often making the mistake of not typing a_{\text{b}}.

However, this extra set of braces doesn't help out when subscriptcorrection is used.
One _could_ make a special addition to the macros when subscriptcorrection is used that would allow
this, but this seem utterly ridiculous. Why, for example, should
a_\text{b} be OK, even though a_\hat{b} isn't !!!!

WaS · Joined: 07 Feb 2006 Posts: 27 Location: Erlangen, Germany

jautschbach · Joined: 17 Mar 2006 Posts: 11

WaS · Joined: 07 Feb 2006 Posts: 27 Location: Erlangen, Germany

Michael Spivak · Joined: 10 Oct 2005 Posts: 52

Perhaps the last post by jautschbach was made having read only WaS' reply, not mine.

I would like to point out that I never said anything about bad style. [In fact, in my opinion, it's crummy style to type extra braces when they are not needed. I always type \hat a, not \hat{a}, although I presume that
LaTeXites would always type the latter.]

My point was, and remains, that it's just a weird happenstance that
the amsmath package defines \text in such a way that a_\text{bc} works
[this wasn't the case with the original definition in amstex.tex], and
that this conflicts with virtually all other situations, like a_\hat{b}, which
don't work even when _ is used in the standard way; similarly, if a user
were to define a new command \mycmd with one argument, then
a_\mycmd{b} wouldn't work either.

If any remarks about style should be added, it should be that whoever added the extra braces to the definition of \text totally obfuscated the issue by creating a special anomalous case that will only serve to confuse everyone unnecessarily!

WaS · Joined: 07 Feb 2006 Posts: 27 Location: Erlangen, Germany

Michael,
the issue is not specific for the \text command: For instance, in standard LaTeX
(without subscript correction) you can also write $a_\mathrm{b}$, whereas
$a_\hat{b}$ would not work. That's why omitting the braces in these cases is
dangerous and is never recommended. Maybe I should have written "sloppy style"
instead. of "bad style".

I suppose the programmers at AMS have defined \text the way they did, because
textual subscripts are often needed, and they wanted \text to behave in the same
way as \mathrm, even though this is neither documented nor advertised in any way.

Michael Spivak · Joined: 10 Oct 2005 Posts: 52

WaS · Joined: 07 Feb 2006 Posts: 27 Location: Erlangen, Germany

good news:

Mike Spivak has found a new implementation for the subscript correction,
which should behave like standard LaTeX, i.e., _\text and _\mathrm should
no longer fail. There will soon be a beta release including this patch.