Date: Tue, 04 Jan 1994 12:37:30 +0000
From: Robin.Fairbairns@cl.cam.ac.uk
Subject: Re: The joys of 2e

You write:
|>         As I am currently the maintainer of changebar I am also
|>         planning to make a new release for LaTeX2e. Any help with that
|>         would of course be appreciated.

As promised, here's a change file for what I've done.

277,288d276
< % \begin{macro}{\atcatcode}
< %    In this file, a lot of control sequences contain {\tt @}-signs.
< %    To be able to do this the category of the {\tt @} has to be `letter'.
< %    Therefore we save the current category code and restore it later.
< %    For this purpose the control sequence \verb=\atcatcode= is used.
< %
< %    \begin{macrocode}
< \chardef\atcatcode=\catcode`\@
< \catcode`\@=11\relax
< %    \end{macrocode}
< % \end{macro}
< %
446,448c434,437
< % \begin{macro}{\driver}
< %    The user can select the specials that should be used by calling
< %    the command {\tt\bsl driver\{}\meta{drivername}{\tt\}}.
---
> % \begin{macro}{package options}
> %    The user can select the specials that should be used by
> %    specifying the driver name as an option to the {\tt\bsl
> %    usepackage} call.
456d444
< %    This command can only be used in the preamble of the document.
458,461c446,448
< %    The argument should be case-insensitive, so it is turned into
< %    a string containing all uppercase characters. To keep some definitions
< %    local, everything is done within a group.
< % \changes{3.1}{2 jul 93}{Removed some spurious spaces}
---
> %    The ambition is that the driver names should be case-insensitive,
> %    but the following code doesn't achieve this: it only permits the
> %    forms given above and their lower-case equivalents.
463,469c450,461
< \def\driver#1{%
<   \bgroup\edef\next{\def\noexpand\tempa{#1}}%
<     \uppercase\expandafter{\next}%
<     \def\LN{DVITOLN03}%
<     \def\DVItoPS{DVITOPS}%
<     \def\DVIPS{DVIPS}%
<     \def\emTeX{EMTEX}%
---
> \DeclareOption{DVItoLN03}{\global\chardef\cb@driver@setup=0\relax}
> \DeclareOption{dvitoln03}{\global\chardef\cb@driver@setup=0\relax}
> \DeclareOption{DVItoPS}{\global\chardef\cb@driver@setup=1\relax}
> \DeclareOption{dvitops}{\global\chardef\cb@driver@setup=1\relax}
> \DeclareOption{DVIps}{\global\chardef\cb@driver@setup=2\relax}
> \DeclareOption{dvips}{\global\chardef\cb@driver@setup=2\relax}
> \DeclareOption{emTeX}{\global\chardef\cb@driver@setup=3\relax}
> \DeclareOption{emtex}{\global\chardef\cb@driver@setup=3\relax}
> \DeclareOption*{\@latexerr%
>       {Unrecognised printer option `\CurrentOption'}%
>       {Allowed options are DVItoLN03, DVItoPS, dvips and emTeX}}
> \ProcessOptions\relax
471,482d462
< %    The choice has to be communicated to the macro \verb=\cb@setup@specials=
< %    that will be called from within \verb=\document=. For this puprose the
< %    control sequence \verb=\cb@driver@setup= is used. It receives a numeric
< %    value using \verb=\chardef=.
< %    \begin{macrocode}
<     \global\chardef\cb@driver@setup=0\relax
<     \ifx\tempa\LN      \global\chardef\cb@driver@setup=0\fi
<     \ifx\tempa\DVItoPS \global\chardef\cb@driver@setup=1\fi
<     \ifx\tempa\DVIPS   \global\chardef\cb@driver@setup=2\fi
<     \ifx\tempa\emTeX   \global\chardef\cb@driver@setup=3\fi
<   \egroup}
< %    \end{macrocode}
485,495d464
< % \begin{macro}{\@preamblecmds}
< %    We add \verb+\driver+ to \verb+\@preamblecmds+, which is a
< %    list of commands to be used only in the preamble of a document.
< %
< %    \begin{macrocode}
< {\def\do{\noexpand\do\noexpand}
<  \xdef\@preamblecmds{\@preamblecmds \do\driver}
< }
< %    \end{macrocode}
< % \end{macro}
< %
1030c999
< % \begin{macro}{\cb@cardr}
---
> % \begin{macro}{\cb@carcdr}
1063c1032
< % \begin{macro}{\document}
---
> % \begin{macro}{\AtBeginDocument}
1065,1067c1034
< %    For these macros we need to modify the \LaTeX-command
< %    \verb=\begin{document}=. The way to this is first save
< %    the old definition of the macro.
---
> %    For these macros we need to add to the LaTeX begin-document hook.
1069c1036
< \let\cb@document=\document
---
> \AtBeginDocument{%
1071d1037
< %    Then it can be redefined including a call to the saved old definition.
1080,1084c1046,1049
< \def\document{\cb@document
<               \cb@setup@specials
<               \cb@positions
<               \immediate\closeout\cb@write
<               \immediate\openin\cb@read=\jobname.cb}
---
>                  \cb@setup@specials
>                  \cb@positions
>                  \immediate\closeout\cb@write
>                  \immediate\openin\cb@read=\jobname.cb}
1088c1053
< % \begin{macro}{\enddocument}
---
> % \begin{macro}{\AtEndDocument}
1090c1055,1059
< %    Does a \verb=\clearpage= to flush rest of document.
---
> %    Does a \verb=\clearpage= to flush rest of document.  (Note that I
> %    believe there is contention in this area: are there in fact
> %    situations in which the end-document hooks need to be called
> %    \emph{before} the final \verb'clearpage'? --- the documentation of
> %    LaTeX itself implies that there are.)
1096,1100c1065
< \let\cb@enddocument\enddocument
< %    \end{macrocode}
< %
< %    \begin{macrocode}
< \def\enddocument{\clearpage\cb@initlist\cb@history
---
> \AtEndDocument{\clearpage\cb@initlist\cb@history
1102c1067
<    \let\cb@barpoint=\cb@checkHistory\cb@enddocument}
---
>    \let\cb@barpoint=\cb@checkHistory}% \cb@enddocument}
1185c1150
<     \cb@footnote{#1}
---
>     \cb@footnote{#1}% this comment removes redundant spaces
1206c1171
<     \cb@mpfootnote{#1}
---
>     \cb@mpfootnote{#1}% this comment removes redundant space
1215,1219d1179
< %    Finally, the category code of {\tt @} is reset to its original
< %    value. The macrospace used by \verb=\atcatcode= is freed.
< %    \begin{macrocode}
< \catcode`\@=\atcatcode \let\atcatcode\relax
< %    \end{macrocode}

Notes:

1) I don't like my implementation of the package options: it doesn't
provide the case flexibility that the present version offers.  Doing it in
a  \DeclareOption* is a trivial modification of the present version.

2) I've posted a comment on the requirement for \clearpage in the end
document hook to LATEX-L (I've cc'ed myself, so if it fails I can provide
a copy direct to you).

3) The comments after executing \cb@[mp]footnote fix a bug that I had
corrected in my own version long ago, but which I appear not to have
posted anywhere...

Robin
--
Robin (Campaign for Real Radio 3) Fairbairns       rf@cl.cam.ac.uk
U of Cambridge Computer Lab, Pembroke St, Cambridge  CB2 3QG, UK

Date: Sun, 06 Feb 1994 10:45:59 +0100
From: Chris Rowley <C.A.Rowley@open.ac.uk>
Subject: RE: a problem with changebar and the ltoutput

Frank wrote:

> Chris & Johannes,
> 
> while recompiling the companion under 2e i ran into a problem with the
> changebar style. this style graps \@makecol and thereby kills the
> reference point of box 255. the result is a page with a lot of stuff
> below the bottom line.
> 
> I'm not completely sure when this happens but the extra line below
> fixes the case in the companion. it didn't happen in 209, so i suppose
> it is due to moving such settings in the output routine.
> 
> not really a bug in the output routine but something to think about.
> 
> the situation was that a large box with a huge depth was at the bottom
> of the page (an example environment). the following redefinition of
> makecol then placed the thing below the the page (when used without
> the boxmaxdepth line).
> 
> \def\@makecol{%
>   \boxmaxdepth\maxdepth  %FMi
>   \setbox\@cclv=\vbox{\cb@resetpoints\cb@beginSaves\unvbox\@cclv}%
>   \gdef\cb@beginSaves{}
>   \global\advance\cb@pagecount by 1%
>   \cb@buildActive\cb@processActive
>   \cb@makecol}
> 
> reason is the kern minus dp output box that is inside the output
> routine somewhere.
If Johannes, or anyone else, is working on things which redefine stuff
in ltoutput, then please:

1. Read the comments in ltoutput.dtx which give some hints about why
things changed and, more importantly, things which I know still need
sttention.  One example is setting of boxmaxdepth.

2. Please send me, or alert me what file to look at, when you redefine
something in another file.  then I can see how you are using things
and thus what the standard case needs to support.

Thanks

Note:  When I get back to looking at ltoutput in detail again, I shall
try to write down some principles for discusssion.

> however, for some reason i was unable to produce this behavior with a
> simple example (nevertheless i'm think that the analysis is basically
> correct) perhaps it was just too late last night
> 
> 
> comments? 
It's always like that with output stuff, simple examples rarely prove
much!

> 
> ps: if chris thinks this is okay can you update the package file? (one
> day)
Assuming that it is only the box being constructed by the \setbox whose
depth should remain small, then I would put the \boxmaxdepth setting inside it
to keep it local; otherwise it may still be in force at times when it
is not wanted, thus:

\def\@makecol{%
%  \boxmaxdepth\maxdepth  %FMi
  \setbox\@cclv \vbox{%
           \cb@resetpoints
	   \cb@beginSaves
	   \unvbox\@cclv
           \boxmaxdepth\maxdepth  %CAR
	              }%
  \gdef\cb@beginSaves{}
  \global\advance\cb@pagecount by 1%
  \cb@buildActive\cb@processActive
  \cb@makecol}

Yes, this does affect the depth of the box inside which it
appears!...really, it does.


chris