[Tex/LaTex] How to resolve clashing definitions of \comment in comment.sty and changes.sty without affecting either’s interface

changescommentsenvironmentsmacrospackage-writing

I'm writing a package that uses both changes (v3.1.2) and comment (v3.8), and I'd like to expose their functionality to consumers of my package without having to modify either one's interface.

The problem is that both changes.sty and comment.sty define \comment. Hence

\documentclass{article}
\usepackage{comment}
\usepackage{changes}
\begin{document}\end{document}

throws the error

Command \comment already defined.

I understand this can be fixed using savesym by (essentially) modifying changes.sty's interface, e.g.

\documentclass{article}
\usepackage{savesym}

\usepackage{changes}
\savesymbol{comment} % rename \comment -> \origcomment
\usepackage{comment}

\begin{document}
\origcomment{pdf comment using 'changes.sty'}
\begin{comment}
code comment using 'comment.sty'
\end{comment}
\end{document}

Alternatively, comment.sty provides \excludecomment for defining custom comment environments, e.g.

\documentclass{article}
\usepackage{savesym}

\usepackage{comment}
\savesymbol{comment}
\excludecomment{mycomment}
\usepackage{changes}

\begin{document}
\comment{pdf comment using 'changes.sty'}
\begin{mycomment}
code comment using 'comment.sty'
\end{mycomment}
\end{document}

but changes.sty breaks the default comment environment.

Is it possible to build a layer in my package which mediates between changes.sty's \comment command and comment.sty's comment environment? The aim is to expose the original interface of each to a consumer of my package.

The two clashing definitions are as follows:

% changes.sty line 791
\newcommand{\comment}[2][\@empty]{%
\setkeys{Changes@comment}{#1}%
\Changes@output%
{comment}%
{\Changes@comment@id}%
{}%
{}%
{#2}%
{\changescommentname}%
{#2}%
}

% comment.sty line 241
\def\excludecomment
 #1{\message{Excluding comment '#1'}%
    \csarg\def{#1}{\endgroup \message{Excluding '#1' comment.}%
        \begingroup
           \DefaultCutFileName \def\ProcessCutFile{}%
           \def\ThisComment####1{}\ProcessComment{#1}}%
    \csarg\def{After#1Comment}{\CloseAndInputCutFile \endgroup}
    \CommentEndDef{#1}}
% ...
% line 292
\excludecomment{comment}

Best Answer

You can via \let save both the \comment-macro which underlies the comment-environment of the comment package and the \comment-macro which comes with the changes-package.

Then you can define your own \comment-macro which "looks" at the definition of \@currenvir.

If \@currenvir = comment, then executing \comment imples calling the macro which underlies the comment-environment.
Otherwise executing \comment implies calling the macro which comes from the changes-package.

Be aware that with the following example I said \includecomment{comment} so that the content of comment-environments results in visual output.

\documentclass{article}

\makeatletter
\newcommand*\UD@CommentName{comment}%
\newcommand\UDsavedcommentenvironment{}%
\newcommand\UDsavedcommentmacro{}%
\usepackage{changes}[2019/01/26 v3.1.2]
\let\UDsavedcommentmacro=\comment
\let\comment=\relax
\usepackage{comment}%[2016/07/16 v3.8]
\includecomment{comment}%
%\excludecomment{comment}%
\let\UDsavedcommentenvironment=\comment
\let\comment=\relax
\newcommand\comment{%
  \ifx\@currenvir\UD@CommentName
    \expandafter\@firstoftwo
  \else
    \expandafter\@secondoftwo
  \fi
  {\UDsavedcommentenvironment}{\UDsavedcommentmacro}%
}%
\makeatother

\begin{document}
\null
\comment{A pdf comment using `changes.sty'.}%
\begin{comment}
A code comment using `comment.sty'.
\end{comment}
\comment{Another pdf comment using `changes.sty'.}%
\begin{comment}
Another code comment using `comment.sty'.
\end{comment}
Some text.
\end{document}

Related Solutions

[Tex/LaTex] Different command definitions with and without optional argument

If you read deep in the LaTeX code you can find lots of examples where this is done. Any command that has optional arguments actually does this already: there are actually two different commands and which one is called depends on whether or not it is called with an optional command. The TeX trick is to use \@ifnextchar[. For example,

\def\mycmd{\@ifnextchar[{\@with}{\@without}}
\def\@with[#1]#2{hello #1, have you met #2?}
\def\@without#1{goodbye #1}

What happens there is that the TeX interpreter meets the \mycmd macro and expands it (swallowing following white-space, of course). The first thing it then encounters is a test on the next character (which, incidentally, is non-destructive - the next character is simply observed, not processed). If it is a square bracket, then it puts \@with in the stream and starts again. \@with is defined to take two arguments, the first of which is surrounded by square brackets. It therefore matches the first optional argument and the next thingy in the TeX stream. If the next character wasn't a square bracket then \@without is put into the stream. This takes one (normal) parameter. But since \@with and \@without are two completely separate commands, they can do whatever they like with the input.

(Note: For commands defined with \newcommand which take an optional parameter then sort-of what happens is that the two commands \@with and \@without expand to the same command, but the optional parameter of \@with gets passed to it as the first parameter whereas in \@without the default value is passed. It's not quite like that, but the difference is more in conciseness of programming than anything else.)

Here's a fully compilable example, including the mysterious and mystical duo \makeatletter and \makeatother:

\documentclass{article}
%\url{https://tex.stackexchange.com/a/314/86}

\makeatletter
\def\mycmd{\@ifnextchar[{\@with}{\@without}}
\def\@with[#1]#2{Hello #1, have you met #2?}
\def\@without#1{Goodbye #1.}
\makeatother

\begin{document}
\mycmd[Polyphemus]{Odysseus}
\mycmd{Circe}
\end{document}

Produces:

Hello Polyphemus, have you met Odysseus? Goodbye Circe.

[Tex/LaTex] Comment out lines without using % and comment enviroment

If your input files all have the same form, that is some lines starting with \\ and an empty line after them, you can do by temporarily redefining \\ and changing the command for inputting the exercises:

\newcommand{\exeinput}[1]{%
  \let\latexdoublebackslash\\
  \ifprolog\let\\\showprolog\else\let\\\hideprolog\fi
  \input{#1}}

\def\showprolog#1\par{%
  \def\\{\par\noindent}% redefine `\\` to end lines
  \par\noindent
  #1% print the lines
  \par % end the last line
  \let\\\latexdoublebackslash}

\def\hideprolog#1\par{\let\\\latexdoublebackslash} % throw away the prolog lines

\newif\ifprolog

So you can say

\exeinput{exercise.435.tex}

and if you set \prologtrue also the prolog lines will be printed.

However this would spectacularly fail if there's not an empty line between the prolog and the main contents, so a special markup is probably the best strategy. For instance, if your exercise files have the form

\startprolog
\\ is the  exercise.435.tex % is the name of the file
\\ 1.84 ex. 127 page notes  % is the source of exercise etc
\stopprolog

\begin{exercise}    
. . . 
\end{exercise}

you can say

\newif\ifprolog
\long\def\startprolog#1\stopprolog{%
  \ifprolog
    \par
    \begingroup
    \let\\\par
    \color{red}\small #1
    \par\medskip
    \endgroup
  \fi}

and so

\input{exercise.435.tex}

will print the exercise without the prolog unless you set \prologtrue. The advantages are that you risk nothing if the prolog is not separated from the contents and that it's possible to reformat freely the prolog when you want to print it (maybe as a marginal note, for instance).

Best Answer

Related Solutions

[Tex/LaTex] Different command definitions with and without optional argument

[Tex/LaTex] Comment out lines without using % and comment enviroment

Related Question