gawk-diffs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[gawk-diffs] [SCM] gawk branch, gawk-4.1-stable, updated. gawk-4.1.0-740


From: Arnold Robbins
Subject: [gawk-diffs] [SCM] gawk branch, gawk-4.1-stable, updated. gawk-4.1.0-740-g6d5c1e1
Date: Sun, 04 Oct 2015 08:04:36 +0000

This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "gawk".

The branch, gawk-4.1-stable has been updated
       via  6d5c1e1d38f71fdfc1794b5b3723a6e476ffddf4 (commit)
      from  a503e996e396e03f819fd0d1710c5e67ed2a59ec (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
http://git.sv.gnu.org/cgit/gawk.git/commit/?id=6d5c1e1d38f71fdfc1794b5b3723a6e476ffddf4

commit 6d5c1e1d38f71fdfc1794b5b3723a6e476ffddf4
Author: Arnold D. Robbins <address@hidden>
Date:   Sun Oct 4 11:04:23 2015 +0300

    Revert update of texinfo.tex.

diff --git a/doc/ChangeLog b/doc/ChangeLog
index ceb2e19..8a0d196 100644
--- a/doc/ChangeLog
+++ b/doc/ChangeLog
@@ -1,3 +1,9 @@
+2015-10-04         Arnold D. Robbins     <address@hidden>
+
+       * texinfo.tex: Revert update. It stopped working. I should learn
+       to test these things. Thanks to Antonio Giovanni Colombo for
+       the report.
+
 2015-10-02         Arnold D. Robbins     <address@hidden>
 
        * gawktexi.in: Note that there is no support for SSL.
diff --git a/doc/texinfo.tex b/doc/texinfo.tex
index 347f863..152e985 100644
--- a/doc/texinfo.tex
+++ b/doc/texinfo.tex
@@ -3,7 +3,7 @@
 % Load plain if necessary, i.e., if running under initex.
 \expandafter\ifx\csname fmtname\endcsname\relax\input plain\fi
 %
-\def\texinfoversion{2015-09-23.01}
+\def\texinfoversion{2015-05-06.11}
 %
 % Copyright 1985, 1986, 1988, 1990, 1991, 1992, 1993, 1994, 1995,
 % 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
@@ -196,6 +196,17 @@
   wide-spread wrap-around
 }
 
+% Margin to add to right of even pages, to left of odd pages.
+\newdimen\bindingoffset
+\newdimen\normaloffset
+\newdimen\pagewidth \newdimen\pageheight
+
+% For a final copy, take out the rectangles
+% that mark overfull boxes (in case you have decided
+% that the text looks ok even though it passes the margin).
+%
+\def\finalout{\overfullrule=0pt }
+
 % Sometimes it is convenient to have everything in the transcript file
 % and nothing on the terminal.  We don't just call \tracingall here,
 % since that produces some useless output on the terminal.  We also make
@@ -240,15 +251,6 @@
 \def\bigbreak{\ifnum\lastpenalty<10000\par\ifdim\lastskip<\bigskipamount
   \removelastskip\penalty-200\bigskip\fi\fi}
 
-% Output routine
-%
-
-% For a final copy, take out the rectangles
-% that mark overfull boxes (in case you have decided
-% that the text looks ok even though it passes the margin).
-%
-\def\finalout{\overfullrule=0pt }
-
 % Do @cropmarks to get crop marks.
 %
 \newif\ifcropmarks
@@ -275,7 +277,6 @@
 % described on page 260 of The TeXbook.  It involves outputting two
 % marks for the sectioning macros, one before the section break, and
 % one after.  I won't pretend I can describe this better than DEK...
-%
 \def\domark{%
   \toks0=\expandafter{\lastchapterdefs}%
   \toks2=\expandafter{\lastsectiondefs}%
@@ -288,9 +289,6 @@
     \noexpand\else \the\toks8             % 2: color marks
   }%
 }
-
-% \gettopheadingmarks, \getbottomheadingmarks - extract needed part of mark.
-%
 % \topmark doesn't work for the very first chapter (after the title
 % page or the contents), so we use \firstmark there -- this gets us
 % the mark with the chapter defs, unless the user sneaks in, e.g.,
@@ -306,74 +304,36 @@
 % Avoid "undefined control sequence" errors.
 \def\lastchapterdefs{}
 \def\lastsectiondefs{}
-\def\lastsection{}
 \def\prevchapterdefs{}
 \def\prevsectiondefs{}
 \def\lastcolordefs{}
 
-% Margin to add to right of even pages, to left of odd pages.
-\newdimen\bindingoffset
-\newdimen\normaloffset
-\newdimen\pagewidth \newdimen\pageheight
-
 % Main output routine.
-%
 \chardef\PAGE = 255
 \output = {\onepageout{\pagecontents\PAGE}}
 
 \newbox\headlinebox
 \newbox\footlinebox
 
-% \onepageout takes a vbox as an argument.
-% \shipout a vbox for a single page, adding an optional header, footer,
-% cropmarks, and footnote.  This also causes index entries for this page
-% to be written to the auxiliary files.
-%
+% \onepageout takes a vbox as an argument.  Note that \pagecontents
+% does insertions, but you have to call it yourself.
 \def\onepageout#1{%
   \ifcropmarks \hoffset=0pt \else \hoffset=\normaloffset \fi
   %
   \ifodd\pageno  \advance\hoffset by \bindingoffset
   \else \advance\hoffset by -\bindingoffset\fi
   %
-  % Common context changes for both heading and footing.
   % Do this outside of the \shipout so @code etc. will be expanded in
   % the headline as they should be, not taken literally (outputting ''code).
   \def\commmonheadfootline{\let\hsize=\pagewidth \texinfochars}
   %
-  % Retrieve the information for the headings from the marks in the page,
-  % and call Plain TeX's \makeheadline and \makefootline, which use the
-  % values in \headline and \footline.
-  %
-  % This is used to check if we are on the first page of a chapter.
-  \ifcase0\topmark\fi
-  \ifx\thischapter\empty
-    % See comment for \gettopheadingmarks
-    \ifcase0\firstmark\fi
-    \let\curchaptername\thischaptername
-    \ifcase1\firstmark\fi
-    \let\prevchaptername\thischaptername
-  \else
-    \let\curchaptername\thischaptername
-    \ifcase1\topmark\fi
-    \let\prevchaptername\thischaptername
-  \fi
-  %
   \ifodd\pageno \getoddheadingmarks \else \getevenheadingmarks \fi
-  \ifodd\pageno \getoddfootingmarks \else \getevenfootingmarks \fi
-  %
-  \ifx\curchaptername\prevchaptername
-  \else
-    % If on the first page of a chapter, clear @thischapter so it
-    % doesn't appear in the headline, because the chapter is already
-    % shown in the chapter heading.
-    \def\thischapter{}%
-  \fi
-  %
   \global\setbox\headlinebox = \vbox{\commmonheadfootline \makeheadline}%
+  %
+  \ifodd\pageno \getoddfootingmarks \else \getevenfootingmarks \fi
   \global\setbox\footlinebox = \vbox{\commmonheadfootline \makefootline}%
   %
   {%
-    % Set context for writing to auxiliary files like index files.
     % Have to do this stuff outside the \shipout because we want it to
     % take effect in \write's, yet the group defined by the \vbox ends
     % before the \shipout runs.
@@ -382,10 +342,10 @@
     \normalturnoffactive  % \ in index entries must not stay \, e.g., if
                % the page break happens to be in the middle of an example.
                % We don't want .vr (or whatever) entries like this:
-               % \entry{{\indexbackslash }acronym}{32}{\code {\acronym}}
+               % \entry{{\tt \indexbackslash }acronym}{32}{\code {\acronym}}
                % "\acronym" won't work when it's read back in;
                % it needs to be
-               % {\code {{\backslashcurfont }acronym}
+               % {\code {{\tt \backslashcurfont }acronym}
     \shipout\vbox{%
       % Do this early so pdf references go to the beginning of the page.
       \ifpdfmakepagedest \pdfdest name{\the\pageno} xyz\fi
@@ -443,7 +403,6 @@
 
 \newinsert\margin \dimen\margin=\maxdimen
 
-% Main part of page, including any footnotes
 \def\pagebody#1{\vbox to\pageheight{\boxmaxdepth=\maxdepth #1}}
 {\catcode`\@ =11
 \gdef\pagecontents#1{\ifvoid\topins\else\unvbox\topins\fi
@@ -466,13 +425,9 @@
 \def\nsbot{\vbox
   {\hrule height\cornerlong depth\cornerthick width\cornerthick}}
 
-
-% Argument parsing
-
 % Parse an argument, then pass it to #1.  The argument is the rest of
 % the input line (except we remove a trailing comment).  #1 should be a
 % macro which expects an ordinary undelimited TeX argument.
-% For example, \def\foo{\parsearg\fooxxx}.
 %
 \def\parsearg{\parseargusing{}}
 \def\parseargusing#1#2{%
@@ -491,11 +446,9 @@
   }%
 }
 
-% First remove any @comment, then any @c comment.  Also remove a @texinfoc
-% comment (see \scanmacro for details).  Pass the result on to \argcheckspaces.
+% First remove any @comment, then any @c comment.
 \def\argremovecomment#1\comment#2\ArgTerm{\argremovec #1\c\ArgTerm}
-\def\argremovec#1\c#2\ArgTerm{\argremovetexinfoc #1\texinfoc\ArgTerm}
-\def\argremovetexinfoc#1\texinfoc#2\ArgTerm{\argcheckspaces#1\^^M\ArgTerm}
+\def\argremovec#1\c#2\ArgTerm{\argcheckspaces#1\^^M\ArgTerm}
 
 % Each occurrence of `\^^M' or `<space>\^^M' is replaced by a single space.
 %
@@ -727,12 +680,6 @@
     \endgraf % Not \par, as it may have been set to \lisppar.
     \global\dimen1 = \prevdepth
   \egroup           % End the \vtop.
-  \addgroupbox
-  \prevdepth = \dimen1
-  \checkinserts
-}
-
-\def\addgroupbox{
   % \dimen0 is the vertical size of the group's box.
   \dimen0 = \ht\groupbox  \advance\dimen0 by \dp\groupbox
   % \dimen2 is how much space is left on the page (more or less).
@@ -745,8 +692,9 @@
     \fi
   \fi
   \box\groupbox
+  \prevdepth = \dimen1
+  \checkinserts
 }
-
 %
 % TeX puts in an \escapechar (i.e., `@') at the beginning of the help
 % message, so this ends up printing address@hidden can only ...'.
@@ -989,20 +937,12 @@ where each line of input produces a line of output.}
 % @c is the same as @comment
 % @ignore ... @end ignore  is another way to write a comment
 %
-\def\comment{\begingroup \catcode`\^^M=\active%
address@hidden \catcode`\{=\other \catcode`\}=\other\commentxxx}%
-
-{\catcode`\^^M=\active%
-\gdef\commentxxx#1^^M{\endgroup%
-\futurelet\nexttoken\commentxxxx}%
-\gdef\commentxxxx{\ifx\nexttoken\aftermacro\expandafter\comment\fi}%
-}
-
-\def\c{\begingroup \catcode`\^^M=\active%
+\def\comment{\begingroup \catcode`\^^M=\other%
 address@hidden \catcode`\{=\other \catcode`\}=\other%
-\cxxx}
-{\catcode`\^^M=\active \gdef\cxxx#1^^M{\endgroup}}
-% See comment in \scanmacro about why the definitions of @c and @comment differ
+\commentxxx}
+{\catcode`\^^M=\other \gdef\commentxxx#1^^M{\endgroup}}
+%
+\let\c=\comment
 
 % @paragraphindent NCHARS
 % We'll use ems for NCHARS, close enough.
@@ -1088,8 +1028,45 @@ where each line of input produces a line of output.}
 % @refill is a no-op.
 \let\refill=\relax
 
-% @setfilename INFO-FILENAME - ignored
-\let\setfilename=\comment
+% If working on a large document in chapters, it is convenient to
+% be able to disable indexing, cross-referencing, and contents, for test runs.
+% This is done with @novalidate (before @setfilename).
+%
+\newif\iflinks \linkstrue % by default we want the aux files.
+\let\novalidate = \linksfalse
+
+% @setfilename is done at the beginning of every texinfo file.
+% So open here the files we need to have open while reading the input.
+% This makes it possible to make a .fmt file for texinfo.
+\def\setfilename{%
+   \fixbackslash  % Turn off hack to swallow `\input texinfo'.
+   \iflinks
+     \tryauxfile
+     % Open the new aux file.  TeX will close it automatically at exit.
+     \immediate\openout\auxfile=\jobname.aux
+   \fi % \openindices needs to do some work in any case.
+   \openindices
+   \let\setfilename=\comment % Ignore extra @setfilename cmds.
+   %
+   % If texinfo.cnf is present on the system, read it.
+   % Useful for site-wide @afourpaper, etc.
+   \openin 1 texinfo.cnf
+   \ifeof 1 \else \input texinfo.cnf \fi
+   \closein 1
+   %
+   \comment % Ignore the actual filename.
+}
+
+% Called from \setfilename.
+%
+\def\openindices{%
+  \newindex{cp}%
+  \newcodeindex{fn}%
+  \newcodeindex{vr}%
+  \newcodeindex{tp}%
+  \newcodeindex{ky}%
+  \newcodeindex{pg}%
+}
 
 % @bye.
 \outer\def\bye{\pagealignmacro\tracingstats=1\ptexend}
@@ -1843,10 +1820,8 @@ end
 % A few fonts for @defun names and args.
 \setfont\defbf\bfshape{10}{\magstep1}{OT1}
 \setfont\deftt\ttshape{10}{\magstep1}{OT1TT}
-\setfont\defsl\slshape{10}{\magstep1}{OT1TT}
 \setfont\defttsl\ttslshape{10}{\magstep1}{OT1TT}
-\def\df{\let\tentt=\deftt \let\tenbf = \defbf
-\let\tenttsl=\defttsl \let\tensl=\defsl \bf}
+\def\df{\let\tentt=\deftt \let\tenbf = \defbf \let\tenttsl=\defttsl \bf}
 
 % Fonts for indices, footnotes, small examples (9pt).
 \def\smallnominalsize{9pt}
@@ -1907,7 +1882,6 @@ end
 % Section fonts (14.4pt).
 \def\secnominalsize{14pt}
 \setfont\secrm\rmbshape{12}{\magstep1}{OT1}
-\setfont\secrmnotbold\rmshape{12}{\magstep1}{OT1}
 \setfont\secit\itbshape{10}{\magstep2}{OT1IT}
 \setfont\secsl\slbshape{10}{\magstep2}{OT1}
 \setfont\sectt\ttbshape{12}{\magstep1}{OT1TT}
@@ -1977,10 +1951,8 @@ end
 % A few fonts for @defun names and args.
 \setfont\defbf\bfshape{10}{\magstephalf}{OT1}
 \setfont\deftt\ttshape{10}{\magstephalf}{OT1TT}
-\setfont\defsl\slshape{10}{\magstephalf}{OT1TT}
 \setfont\defttsl\ttslshape{10}{\magstephalf}{OT1TT}
-\def\df{\let\tentt=\deftt \let\tenbf = \defbf
-\let\tensl=\defsl \let\tenttsl=\defttsl \bf}
+\def\df{\let\tentt=\deftt \let\tenbf = \defbf \let\tenttsl=\defttsl \bf}
 
 % Fonts for indices, footnotes, small examples (9pt).
 \def\smallnominalsize{9pt}
@@ -2799,6 +2771,11 @@ end
 %
 \def\dmn#1{\thinspace #1}
 
+% @l was never documented to mean ``switch to the Lisp font'',
+% and it is not used as such in any manual I can find.  We need it for
+% Polish suppressed-l.  --karl, 22sep96.
+%\def\l#1{{\li #1}\null}
+
 % @acronym for "FBI", "NATO", and the like.
 % We print this one point size smaller, since it's intended for
 % all-uppercase.
@@ -2902,6 +2879,9 @@ end
 \def\sup{\ifmmode \expandafter\ptexsp \else \expandafter\finishsup\fi}
 \def\finishsup#1{$\ptexsp{\hbox{\selectfonts\lllsize #1}}$}%
 
+% ctrl is no longer a Texinfo command, but leave this definition for fun.
+\def\ctrl #1{{\tt \rawbackslash \hat}#1}
+
 % @inlinefmt{FMTNAME,PROCESSED-TEXT} and @inlineraw{FMTNAME,RAW-TEXT}.
 % Ignore unless FMTNAME == tex; then it is like @iftex and @tex,
 % except specified as a normal braced arg, so no newlines to worry about.
@@ -3215,15 +3195,8 @@ end
 \def\Eogonek{{\ecfont \char"86}}\def\macrocharE{E}
 \def\eogonek{{\ecfont \char"A6}}\def\macrochare{e}
 %
-% Use the European Computer Modern fonts (cm-super in outline format)
-% for non-CM glyphs.  That is ec* for regular text and tc* for the text
-% companion symbols (LaTeX TS1 encoding).  Both are part of the ec
-% package and follow the same conventions.
-% 
-\def\ecfont{\etcfont{e}}
-\def\tcfont{\etcfont{t}}
-%
-\def\etcfont#1{%
+% Use the ec* fonts (cm-super in outline format) for non-CM glyphs.
+\def\ecfont{%
   % We can't distinguish serif/sans and italic/slanted, but this
   % is used for crude hacks anyway (like adding French and German
   % quotes to documents typeset with CM, where we lose kerning), so
@@ -3232,14 +3205,14 @@ end
   \edef\nominalsize{\csname\curfontsize nominalsize\endcsname}%
   \ifmonospace
     % typewriter:
-    \font\thisecfont = #1ctt\ecsize \space at \nominalsize
+    \font\thisecfont = ectt\ecsize \space at \nominalsize
   \else
     \ifx\curfontstyle\bfstylename
       % bold:
-      \font\thisecfont = #1cb\ifusingit{i}{x}\ecsize \space at \nominalsize
+      \font\thisecfont = ecb\ifusingit{i}{x}\ecsize \space at \nominalsize
     \else
       % regular:
-      \font\thisecfont = #1c\ifusingit{ti}{rm}\ecsize \space at \nominalsize
+      \font\thisecfont = ec\ifusingit{ti}{rm}\ecsize \space at \nominalsize
     \fi
   \fi
   \thisecfont
@@ -3409,7 +3382,7 @@ end
 \newtoks\evenfootline    % footline on even pages
 \newtoks\oddfootline     % footline on odd pages
 
-% Now make \makeheadline and \makefootline in Plain TeX use those variables
+% Now make TeX use those variables
 \headline={{\textfonts\rm \ifodd\pageno \the\oddheadline
                             \else \the\evenheadline \fi}}
 \footline={{\textfonts\rm \ifodd\pageno \the\oddfootline
@@ -3465,10 +3438,6 @@ end
 % @everyheadingmarks
 % @everyfootingmarks
 
-% These define \getoddheadingmarks, \getevenheadingmarks,
-% \getoddfootingmarks, and \getevenfootingmarks, each to one of
-% \gettopheadingmarks, \getbottomheadingmarks.
-%
 \def\evenheadingmarks{\headingmarks{even}{heading}}
 \def\oddheadingmarks{\headingmarks{odd}{heading}}
 \def\evenfootingmarks{\headingmarks{even}{footing}}
@@ -3758,12 +3727,7 @@ end
    \noindent
    \hbox to 0pt{\hss \itemcontents \kern\itemmargin}%
    %
-   \ifinner\else
-     \vadjust{\penalty 1200}% not good to break after first line of item.
-   \fi
-   % We can be in inner vertical mode in a footnote, although an
-   % @itemize looks awful there.
-  }%
+   \vadjust{\penalty 1200}}% not good to break after first line of item.
   \flushcr
 }
 
@@ -4385,16 +4349,19 @@ end
 % except not \outer, so it can be used within macros and \if's.
 \edef\newwrite{\makecsname{ptexnewwrite}}
 
-% \newindex {foo} defines an index named IX.
-% It automatically defines \IXindex such that
-% \IXindex ...rest of line... puts an entry in the index IX.
-% It also defines \IXindfile to be the number of the output channel for
-% the file that accumulates this index.  The file's extension is IX.
+% \newindex {foo} defines an index named foo.
+% It automatically defines \fooindex such that
+% \fooindex ...rest of line... puts an entry in the index foo.
+% It also defines \fooindfile to be the number of the output channel for
+% the file that accumulates this index.  The file's extension is foo.
 % The name of an index should be no more than 2 characters long
 % for the sake of vms.
 %
 \def\newindex#1{%
-  \expandafter\chardef\csname#1indfile\endcsname=0
+  \iflinks
+    \expandafter\newwrite \csname#1indfile\endcsname
+    \openout \csname#1indfile\endcsname \jobname.#1 % Open the file
+  \fi
   \expandafter\xdef\csname#1index\endcsname{%     % Define @#1index
     \noexpand\doindex{#1}}
 }
@@ -4408,19 +4375,14 @@ end
 \def\defcodeindex{\parsearg\newcodeindex}
 %
 \def\newcodeindex#1{%
-  \expandafter\chardef\csname#1indfile\endcsname=0
+  \iflinks
+    \expandafter\newwrite \csname#1indfile\endcsname
+    \openout \csname#1indfile\endcsname \jobname.#1
+  \fi
   \expandafter\xdef\csname#1index\endcsname{%
     \noexpand\docodeindex{#1}}%
 }
 
-% The default indices:
-\newindex{cp}%      concepts,
-\newcodeindex{fn}%  functions,
-\newcodeindex{vr}%  variables,
-\newcodeindex{tp}%  types,
-\newcodeindex{ky}%  keys
-\newcodeindex{pg}%  and programs.
-
 
 % @synindex foo bar    makes index foo feed into index bar.
 % Do this instead of @defindex foo if you don't want it as a separate index.
@@ -4449,19 +4411,26 @@ end
   \expandafter\xdef\csname#2index\endcsname{\noexpand#1{#3}}%
 }
 
-% Define \doindex, the driver for all index macros.
+% Define \doindex, the driver for all \fooindex macros.
 % Argument #1 is generated by the calling \fooindex macro,
-% and it the two-letter name of the index.
+%  and it is "foo", the name of the index.
 
-\def\doindex#1{\edef\indexname{#1}\parsearg\doindexxxx}
-\def\doindexxxx #1{\doind{\indexname}{#1}}
+% \doindex just uses \parsearg; it calls \doind for the actual work.
+% This is because \doind is more useful to call from other macros.
+
+% There is also \dosubind {index}{topic}{subtopic}
+% which makes an entry in a two-level index such as the operation index.
+
+\def\doindex#1{\edef\indexname{#1}\parsearg\singleindexer}
+\def\singleindexer #1{\doind{\indexname}{#1}}
 
 % like the previous two, but they put @code around the argument.
-\def\docodeindex#1{\edef\indexname{#1}\parsearg\docodeindexxxx}
-\def\docodeindexxxx #1{\doind{\indexname}{\code{#1}}}
+\def\docodeindex#1{\edef\indexname{#1}\parsearg\singlecodeindexer}
+\def\singlecodeindexer #1{\doind{\indexname}{\code{#1}}}
 
-% Used when writing an index entry out to an index file, to prevent
-% expansion of Texinfo commands that can appear in an index entry.
+% Take care of Texinfo commands that can appear in an index entry.
+% Since there are some commands we want to expand, and others we don't,
+% we have to laboriously prevent expansion for those that we don't.
 %
 \def\indexdummies{%
   \escapechar = `\\     % use backslash in output files.
@@ -4477,6 +4446,31 @@ end
   \def\{{{\tt\char123}}%
   \def\}{{\tt\char125}}%
   %
+  % I don't entirely understand this, but when an index entry is
+  % generated from a macro call, the \endinput which \scanmacro inserts
+  % causes processing to be prematurely terminated.  This is,
+  % apparently, because \indexsorttmp is fully expanded, and \endinput
+  % is an expandable command.  The redefinition below makes \endinput
+  % disappear altogether for that purpose -- although logging shows that
+  % processing continues to some further point.  On the other hand, it
+  % seems \endinput does not hurt in the printed index arg, since that
+  % is still getting written without apparent harm.
+  %
+  % Sample source (mac-idx3.tex, reported by Graham Percival to
+  % help-texinfo, 22may06):
+  % @macro funindex {WORD}
+  % @findex xyz
+  % @end macro
+  % ...
+  % @funindex commtest
+  % This is not enough to reproduce the bug, but it gives the flavor.
+  %
+  % Sample whatsit resulting:
+  % address@hidden@folio address@hidden address@hidden }}}
+  %
+  % So:
+  \let\endinput = \empty
+  %
   % Do the redefinitions.
   \commondummies
 }
@@ -4683,39 +4677,11 @@ end
   % that we make for arg2 (see \parsemargdef ff.).  We want all this to be
   % expanded for the sake of the index, so we end up just seeing "bar".
   \let\xeatspaces = \eatspaces
-  \let\xprocessmacroarg\eatspaces
 }
 
 % For testing: output @{ and @} in index sort strings as \{ and \}.
 \newif\ifusebracesinindexes
 
-\let\indexlbrace\relax
-\let\indexrbrace\relax
-
address@hidden
-\catcode`\\=13
-  @address@hidden@def\{}}
-}
-
-{
-\catcode`\<=13
-\catcode`\-=13
-  \gdef\indexnonalnumdisappear{%
-    \backslashdisappear
-    \def-{}%
-    \def<{}%
-    address@hidden
-  }
-
-  \gdef\indexnonalnumreappear{%
-    \useindexbackslash
-    \let-\normaldash
-    \let<\normalless
-    address@hidden@}%
-  }
-}
-
-
 % \indexnofonts is used when outputting the strings to sort the index
 % by, and when constructing control sequence names.  It eliminates all
 % control sequences and just writes whatever the best ASCII sort string
@@ -4741,10 +4707,16 @@ end
   \def\_{\normalunderscore}%
   \def\-{}% @- shouldn't affect sorting
   %
-  % Unfortunately, texindex < 6.0 is not prepared to handle braces in the
-  % content at all, so these won't be sorted in ASCII order.
-  \def\lbracechar{{\indexlbrace}}%
-  \def\rbracechar{{\indexrbrace}}%
+  % Unfortunately, texindex is not prepared to handle braces in the
+  % content at all.  So for index sorting, we map @{ and @} to strings
+  % starting with |, since that ASCII character is between ASCII { and }.
+  \ifusebracesinindexes
+    \def\lbracechar{\lbracecmd}%
+    \def\rbracechar{\rbracecmd}%
+  \else
+    \def\lbracechar{|a}%
+    \def\rbracechar{|b}%
+  \fi
   \let\{=\lbracechar
   \let\}=\rbracechar
   %
@@ -4756,7 +4728,7 @@ end
   \def\L{L}%
   \def\OE{OE}%
   \def\O{O}%
-  \def\TH{TH}%
+  \def\TH{ZZZ}%
   \def\aa{aa}%
   \def\ae{ae}%
   \def\dh{dzz}%
@@ -4768,7 +4740,7 @@ end
   \def\o{o}%
   \def\questiondown{?}%
   \def\ss{ss}%
-  \def\th{th}%
+  \def\th{zzz}%
   %
   \def\LaTeX{LaTeX}%
   \def\TeX{TeX}%
@@ -4826,17 +4798,14 @@ end
 {\catcode`\`=\active
  \gdef\indexlquoteignore{\let`=\empty}}
 
+\let\indexbackslash=0  %overridden during \printindex.
 \let\SETmarginindex=\relax % put index entries in margin (undocumented)?
 
 % Most index entries go through here, but \dosubind is the general case.
 % #1 is the index name, #2 is the entry text.
 \def\doind#1#2{\dosubind{#1}{#2}{}}
 
-% There is also \dosubind {index}{topic}{subtopic}
-% which makes an entry in a two-level index such as the operation index.
-% TODO: Two-level index?  Operation index?
-
-% Workhorse for all indexes.
+% Workhorse for all \fooindexes.
 % #1 is name of index, #2 is stuff to put there, #3 is subentry --
 % empty if called from \doind, as we usually are (the main exception
 % is with most defuns, which call us directly).
@@ -4844,7 +4813,6 @@ end
 \def\dosubind#1#2#3{%
   \iflinks
   {%
-    \requireopenindexfile{#1}%
     % Store the main index entry text (including the third arg).
     \toks0 = {#2}%
     % If third arg is present, precede it with a space.
@@ -4860,49 +4828,7 @@ end
   \fi
 }
 
-% Check if an index file has been opened, and if not, open it.
-\def\requireopenindexfile#1{%
-\ifnum\csname #1indfile\endcsname=0
-  \expandafter\newwrite \csname#1indfile\endcsname
-  \edef\suffix{#1}%
-  % A .fls suffix would conflict with the file extension for the output
-  % of -recorder, so use .f1s instead.
-  \ifx\suffix\indexisfl\def\suffix{f1}\fi
-  % Open the file
-  \immediate\openout\csname#1indfile\endcsname \jobname.\suffix
-  % Using \immediate here prevents an object entering into the current box,
-  % which could confound checks such as those in \safewhatsit for preceding
-  % skips.
-\fi}
-\def\indexisfl{fl}
-
-% Output \ as {\indexbackslash}, because \ is an escape character in
-% the index files.
-\let\indexbackslash=\relax
address@hidden \catcode`\\=\active
-  @address@hidden@address@hidden
-}
-
-% Definition for writing index entry text.
-\def\sortas#1{\ignorespaces}%
-
-% Definition for writing index entry sort key.  Should occur at the at
-% the beginning of the index entry, like
-%     @cindex @sortas{september} \september
-% The \ignorespaces takes care of following space, but there's no way
-% to remove space before it.
-{
-\catcode`\-=13
-\gdef\indexwritesortas{%
-  \begingroup
-  \indexnonalnumreappear
-  \indexwritesortasxxx}
-\gdef\indexwritesortasxxx#1{%
-  \xdef\indexsortkey{#1}\endgroup}
-}
-
-
-% Write the entry in \toks0 to the index file.
+% Write the entry in \toks0 to the index file:
 %
 \def\dosubindwrite{%
   % Put the index entry in the margin if desired.
@@ -4912,20 +4838,14 @@ end
   %
   % Remember, we are within a group.
   \indexdummies % Must do this here, since \bf, etc expand at this stage
-  \useindexbackslash % \indexbackslash isn't defined now so it will be output 
-                     % as is; and it will print as backslash.
-  % Get the string to sort by, by processing the index entry with all
-  % font commands turned off.
+  \def\backslashcurfont{\indexbackslash}% \indexbackslash isn't defined now
+      % so it will be output as is; and it will print as backslash.
+  %
+  % Process the index entry with all font commands turned off, to
+  % get the string to sort by.
   {\indexnofonts
-   \xdef\indexsortkey{}%
-   \let\sortas=\indexwritesortas
-   \indexnonalnumdisappear
-   \edef\temp{\the\toks0}%
-   \setbox\dummybox = \hbox{\temp}% Make sure to execute any \sortas
-   \ifx\indexsortkey\empty
-     \xdef\indexsortkey{\temp}%
-     \ifx\indexsortkey\empty\xdef\indexsortkey{ }\fi
-   \fi
+   \edef\temp{\the\toks0}% need full expansion
+   \xdef\indexsorttmp{\temp}%
   }%
   %
   % Set up the complete index entry, with both the sort key and
@@ -4935,11 +4855,10 @@ end
   % sorted result.
   \edef\temp{%
     \write\writeto{%
-      \string\entry{\indexsortkey}{\noexpand\folio}{\the\toks0}}%
+      \string\entry{\indexsorttmp}{\noexpand\folio}{\the\toks0}}%
   }%
   \temp
 }
-\newbox\dummybox % used above
 
 % Take care of unwanted page breaks/skips around a whatsit:
 %
@@ -5065,9 +4984,7 @@ end
   % as its first line, TeX doesn't complain about mismatched braces
   % (because it thinks @} is a control sequence).
   \catcode`\@ = 11
-  % See comment in \requireopenindexfile.
-  \def\indexname{#1}\ifx\indexname\indexisfl\def\indexname{f1}\fi
-  \openin 1 \jobname.\indexname s
+  \openin 1 \jobname.#1s
   \ifeof 1
     % \enddoublecolumns gets confused if there is no text in the index,
     % and it loses the chapter title and the aux file entries for the
@@ -5075,96 +4992,43 @@ end
     % there is some text.
     \putwordIndexNonexistent
   \else
-    \catcode`\\ = 0
-    \escapechar = `\\
     %
     % If the index file exists but is empty, then \openin leaves \ifeof
     % false.  We have to make TeX try to read something from the file, so
     % it can discover if there is anything in it.
-    \read 1 to \thisline
+    \read 1 to \temp
     \ifeof 1
       \putwordIndexIsEmpty
     \else
       % Index files are almost Texinfo source, but we use \ as the escape
       % character.  It would be better to use @, but that's too big a change
       % to make right now.
-      \def\indexbackslash{\ttbackslash}%
-      \let\indexlbrace\{   % Likewise, set these sequences for braces
-      \let\indexrbrace\}   % used in the sort key.
+      \def\indexbackslash{\backslashcurfont}%
+      \catcode`\\ = 0
+      \escapechar = `\\
       \begindoublecolumns
-      \let\entryorphanpenalty=\indexorphanpenalty
-      %
-      % Read input from the index file line by line.
-      \loopdo
-        \ifeof1
-          \let\firsttoken\relax
-        \else
-          \read 1 to \nextline
-          \edef\act{\gdef\noexpand\firsttoken{\getfirsttoken\nextline}}%
-          \act
-        \fi
-        \thisline
-        %
-        \ifeof1\else
-        \let\thisline\nextline
-      \repeat
-      %%
+      \input \jobname.#1s
       \enddoublecolumns
     \fi
   \fi
   \closein 1
 \endgroup}
 
-\def\getfirsttoken#1{\expandafter\getfirsttokenx#1\endfirsttoken}
-\long\def\getfirsttokenx#1#2\endfirsttoken{\noexpand#1}
-
-\def\loopdo#1\repeat{\def\body{#1}\loopdoxxx}
-\def\loopdoxxx{\let\next=\relax\body\let\next=\loopdoxxx\fi\next}
-
 % These macros are used by the sorted index file itself.
 % Change them to control the appearance of the index.
 
-{\catcode`\/=13 \catcode`\-=13 \catcode`\^=13 \catcode`\~=13 \catcode`\_=13
-\catcode`\|=13 \catcode`\<=13 \catcode`\>=13 \catcode`\+=13 \catcode`\"=13
-\catcode`\$=3
-\gdef\initialglyphs{%
-  % Some changes for non-alphabetic characters.  Using the glyphs from the
-  % math fonts looks more consistent than the typewriter font used elsewhere
-  % for these characters.
-  \def\indexbackslash{\math{\backslash}}%
-  \let\\=\indexbackslash
-  %
-  % Can't get bold backslash so don't use bold forward slash
-  \catcode`\/=13
-  \def/{{\secrmnotbold \normalslash}}%
-  \def-{{\normaldash\normaldash}}% en dash `--'
-  \def^{{\chapbf \normalcaret}}%
-  \def~{{\chapbf \normaltilde}}%
-  \def\_{%
-     \leavevmode \kern.07em \vbox{\hrule width.3em height.1ex}\kern .07em }%
-  \def|{$\vert$}%
-  \def<{$\less$}%
-  \def>{$\gtr$}%
-  \def+{$\normalplus$}%
-}}
-
-\def\initial{%
-  \bgroup
-  \initialglyphs
-  \initialx
-}
-
-\def\initialx#1{%
+\def\initial#1{{%
+  % Some minor font changes for the special characters.
+  \let\tentt=\sectt \let\tt=\sectt \let\sf=\sectt
+  %
   % Remove any glue we may have, we'll be inserting our own.
   \removelastskip
   %
   % We like breaks before the index initials, so insert a bonus.
-  % The glue before the bonus allows a little bit of space at the
-  % bottom of a column to reduce an increase in inter-line spacing.
   \nobreak
-  \vskip 0pt plus 5\baselineskip
-  \penalty -300 
-  \vskip 0pt plus -5\baselineskip
+  \vskip 0pt plus 3\baselineskip
+  \penalty 0
+  \vskip 0pt plus -3\baselineskip
   %
   % Typeset the initial.  Making this add up to a whole number of
   % baselineskips increases the chance of the dots lining up from column
@@ -5172,24 +5036,24 @@ end
   % we need before each entry, but it's better.
   %
   % No shrink because it confuses \balancecolumns.
-  \vskip 1.67\baselineskip plus 1\baselineskip
-  \leftline{\secfonts \kern-0.05em \secbf #1}%
-  % \secfonts is inside the argument of \leftline so that the change of
-  % \baselineskip will not affect any glue inserted before the vbox that
-  % \leftline creates.
+  \vskip 1.67\baselineskip plus .5\baselineskip
+  \leftline{\secbf #1}%
   % Do our best not to break after the initial.
   \nobreak
   \vskip .33\baselineskip plus .1\baselineskip
-  \egroup % \initialglyphs
-}
-
-\newdimen\entryrightmargin
-\entryrightmargin=0pt
+}}
 
 % \entry typesets a paragraph consisting of the text (#1), dot leaders, and
 % then page number (#2) flushed to the right margin.  It is used for index
 % and table of contents entries.  The paragraph is indented by \leftskip.
 %
+% A straightforward implementation would start like this:
+%      \def\entry#1#2{...
+% But this freezes the catcodes in the argument, and can cause problems to
+% @code, which sets - active.  This problem was fixed by a kludge---
+% ``-'' was active throughout whole index, but this isn't really right.
+% The right solution is to prevent \entry from swallowing the whole text.
+%                                 --kasal, 21nov03
 \def\entry{%
   \begingroup
     %
@@ -5197,46 +5061,38 @@ end
     % affect previous text.
     \par
     %
+    % Do not fill out the last line with white space.
+    \parfillskip = 0in
+    %
     % No extra space above this paragraph.
     \parskip = 0in
     %
-    % When reading the text of entry, convert explicit line breaks
-    % from @* into spaces.  The user might give these in long section
-    % titles, for instance.
-    \def\*{\unskip\space\ignorespaces}%
-    \def\entrybreak{\hfil\break}%
-    %
-    % A bit of stretch before each entry for the benefit of balancing
-    % columns.
-    \vskip 0pt plus0.5pt
-    %
-    % Badness calculation for paragraph affected by -
-    %  How much \indexdotfill is stretched, or how much \parfillskip is shrunk
-    %  Number of lines (\linepenalty)
-    %  
     % Do not prefer a separate line ending with a hyphen to fewer lines.
     \finalhyphendemerits = 0
     %
-    % Word spacing - no stretch
-    \spaceskip=\fontdimen2\font minus \fontdimen4\font
+    % \hangindent is only relevant when the entry text and page number
+    % don't both fit on one line.  In that case, bob suggests starting the
+    % dots pretty far over on the line.  Unfortunately, a large
+    % indentation looks wrong when the entry text itself is broken across
+    % lines.  So we use a small indentation and put up with long leaders.
+    %
+    % \hangafter is reset to 1 (which is the value we want) at the start
+    % of each paragraph, so we need not do anything with that.
+    \hangindent = 2em
     %
-    \linepenalty=1000  % Discourage line breaks.
-    \hyphenpenalty=5000  % Discourage hyphenation.
+    % When the entry text needs to be broken, just fill out the first line
+    % with blank space.
+    \rightskip = 0pt plus1fil
     %
-    % Ragged right margin, but not for the last line with the leaders in it.
-    % When an index entry spans lines, this stretch competes with the stretch
-    % in \indexdotfill to determine how the line will be split.
-    \rightskip=\entryrightmargin
-    \advance\rightskip by 0pt plus .6\hsize
+    % A bit of stretch before each entry for the benefit of balancing
+    % columns.
+    \vskip 0pt plus1pt
     %
-    % \parfillskip is at the end of the line with the page number
-    \parfillskip=0pt
-    % Cancel the \rightskip stretch
-    \advance \parfillskip by 0pt plus -.6\hsize
-    % Determine how far we can stretch into the margin.
-    % This allows, e.g., "Appendix H  GNU Free Documentation License" to fit
-    % on one line.
-    \advance \parfillskip by 0pt minus .6\entryrightmargin
+    % When reading the text of entry, convert explicit line breaks
+    % from @* into spaces.  The user might give these in long section
+    % titles, for instance.
+    \def\*{\unskip\space\ignorespaces}%
+    \def\entrybreak{\hfil\break}%
     %
     % Swallow the left brace of the text (first parameter):
     \afterassignment\doentry
@@ -5244,8 +5100,6 @@ end
 }
 \def\entrybreak{\unskip\space\ignorespaces}%
 \def\doentry{%
-    % Save the text of the entry in a \vtop.
-    \global\setbox\entryindexbox=\vtop\bgroup
     \bgroup % Instead of the swallowed brace.
       \noindent
       \aftergroup\finishentry
@@ -5262,70 +5116,29 @@ end
       \ %
     \else
       %
+      % If we must, put the page number on a line of its own, and fill out
+      % this line with blank space.  (The \hfil is overwhelmed with the
+      % fill leaders glue in \indexdotfill if the page number does fit.)
+      \hfil\penalty50
       \null\nobreak\indexdotfill % Have leaders before the page number.
       %
+      % The `\ ' here is removed by the implicit \unskip that TeX does as
+      % part of (the primitive) \par.  Without it, a spurious underfull
+      % \hbox ensues.
       \ifpdf
-        \pdfgettoks#1.%
-        \hskip\skip\thinshrinkable\the\toksA
+       \pdfgettoks#1.%
+       \ \the\toksA
       \else
-        \hskip\skip\thinshrinkable #1%
+       \ #1%
       \fi
     \fi
-    % Parameters for formatting this paragraph, reset for each paragraph.
-    %
-    % \hangindent is only relevant when the entry text and page number
-    % don't both fit on one line.  In that case, bob suggests starting the
-    % dots pretty far over on the line.  Unfortunately, a large
-    % indentation looks wrong when the entry text itself is broken across
-    % lines.  So we use a small indentation and put up with long leaders.
-    %
-    \hangafter = 1
-    \hangindent = 1em
     \par
-    \egroup % The \vtop
   \endgroup
-  % delay text of entry until after penalty
-  \bgroup\aftergroup\insertindexentrybox
-  \entryorphanpenalty
-}
-
-\newskip\thinshrinkable
-\skip\thinshrinkable=.15em minus .15em
-
-\newbox\entryindexbox
-\def\insertindexentrybox{%
-\lineskip=.8ex plus .6ex % This comes into effect when the \vtop has a large 
-                         % depth due to the paragraph in it having several 
-                         % lines.
-\box\entryindexbox}
-
-% Default is no penalty
-\let\entryorphanpenalty\egroup
-
-% Used from \printindex.  \firsttoken should be the first token
-% after the \entry.  If it's not another \entry, we are at the last
-% line of a group of index entries, so insert a penalty to discourage
-% orphaned index entries.
-\long\def\indexorphanpenalty{%
-  \def\isentry{\entry}%
-  \ifx\firsttoken\isentry
-  \else
-    \unskip\penalty 9000
-    % The \unskip here stops breaking before the glue.  It relies on the
-    % \vskip above being there, otherwise there is an error
-    % "You can't use `\unskip' in vertical mode".  There has to be glue
-    % in the current vertical list that hasn't been added to the
-    % "current page".  See Chapter 24 of the TeXbook.  This contradicts
-    % Section 8.3.7 in "TeX by Topic," though.
-  \fi
-  \egroup % now comes the box added with \aftergroup
 }
 
 % Like plain.tex's \dotfill, except uses up at least 1 em.
-% Using a finite stretch encourages several words to appear on a second line
-% if the entry is broken.
 \def\indexdotfill{\cleaders
-  \hbox{$\mathsurround=0pt \mkern1.5mu.\mkern1.5mu$}\hskip 1em plus .4\hsize}
+  \hbox{$\mathsurround=0pt \mkern1.5mu.\mkern1.5mu$}\hskip 1em plus 1fill}
 
 \def\primary #1{\line{#1\hfil}}
 
@@ -5352,9 +5165,6 @@ end
 \newbox\partialpage
 \newdimen\doublecolumnhsize
 
-\newtoks\savedtopmark % Used in \begindoublecolumns
-\newtoks\savedfirstmark
-
 \def\begindoublecolumns{\begingroup % ended by \enddoublecolumns
   % Grab any single-column material above us.
   \output = {%
@@ -5375,23 +5185,9 @@ end
       \unvbox\PAGE
       \kern-\topskip \kern\baselineskip
     }%
-    % Save \topmark and \firstmark
-    \global\savedtopmark=\expandafter{\topmark}%
-    \global\savedfirstmark=\expandafter{\firstmark}%
   }%
   \eject % run that output routine to set \partialpage
   %
-  % We recover the two marks that the last output routine saved in order
-  % to propagate the information in marks added around a chapter heading,
-  % which could be otherwise be lost by the time the final page is output.
-  %
-  \mark{\the\savedtopmark}% Only mark in page passed to following \output.
-  \output = {%
-    \setbox0=\box\PAGE % clear box 255
-  }abc\eject
-  %
-  \mark{\the\savedfirstmark}%
-  %
   % Use the double-column output routine for subsequent pages.
   \output = {\doublecolumnout}%
   %
@@ -5422,7 +5218,7 @@ end
 }
 
 % The double-column output routine for all double-column pages except
-% the last, which is done by \balancecolumns.
+% the last.
 %
 \def\doublecolumnout{%
   \splittopskip=\topskip \splitmaxdepth=\maxdepth
@@ -5504,47 +5300,28 @@ end
   \pagegoal = \vsize
 }
 %
-% Only called for the last of the double column material.  \doublecolumnout 
-% does the others.
+% Called at the end of the double column material.
 \def\balancecolumns{%
   \setbox0 = \vbox{\unvbox255}% like \box255 but more efficient, see p.120.
   \dimen@ = \ht0
   \advance\dimen@ by \topskip
   \advance\dimen@ by-\baselineskip
-  \ifdim\dimen@<14\baselineskip
-    % Don't split a short final column in two.
-    \setbox2=\vbox{}%
-  \else
-    \divide\dimen@ by 2 % target to split to
-    address@hidden = \dimen@
-    \splittopskip = \topskip
-    % Loop until the second column is no higher than the first
-    {%
-      \vbadness = 10000
-      \loop
-        \global\setbox3 = \copy0
-        \global\setbox1 = \vsplit3 to \dimen@
-        % Remove glue from bottom of first column to
-        % make sure it is higher than the second.
-        \global\setbox1 = \vbox{\unvbox1\unpenalty\unskip}%
-      \ifdim\ht3>\ht1
-        \global\advance\dimen@ by 1pt
-      \repeat
-    }%
-    address@hidden by 4
-    address@hidden by 5
-    \ifdim\ht3<address@hidden
-      % Column heights are too different, so don't make their bottoms
-      % flush with each other.  The glue at the end of the second column
-      % allows a second column to stretch, reducing the difference in
-      % height between the two.
-      \setbox0=\vbox address@hidden
-      \setbox2=\vbox address@hidden 0pt plus 0.3\ht0}%
-    \else
-      \setbox0=\vbox address@hidden
-      \setbox2=\vbox address@hidden
-    \fi
-  \fi
+  \divide\dimen@ by 2 % target to split to
+  %debug\message{final 2-column material height=\the\ht0, address@hidden
+  \splittopskip = \topskip
+  % Loop until we get a decent breakpoint.
+  {%
+    \vbadness = 10000
+    \loop
+      \global\setbox3 = \copy0
+      \global\setbox1 = \vsplit3 to \dimen@
+    \ifdim\ht3>\dimen@
+      \global\advance\dimen@ by 1pt
+    \repeat
+  }%
+  %debug\message{split to \the\dimen@, column heights: \the\ht1, \the\ht3.}%
+  \setbox0=\vbox address@hidden
+  \setbox2=\vbox address@hidden
   %
   \pagesofar
 }
@@ -5565,9 +5342,6 @@ end
     \let\lastnode=\empty      % no node to associate with
     \writetocentry{part}{#1}{}% but put it in the toc
     \headingsoff              % no headline or footline on the part page
-    % This outputs a mark at the end of the page that clears \thischapter
-    % and \thissection, like is done in \startcontents.
-    \chapmacro{}{Yomitfromtoc}{}%
     \chapoddpage
   \endgroup
 }
@@ -5812,6 +5586,9 @@ end
 
 % @centerchap is like @unnumbered, but the heading is centered.
 \outer\parseargdef\centerchap{%
+  % Well, we could do the following in a group, but that would break
+  % an assumption that \chapmacro is called at the outermost level.
+  % Thus we are safer this way:                --kasal, 24feb04
   \let\centerparametersmaybe = \centerparameters
   \unnmhead0{#1}%
   \let\centerparametersmaybe = \relax
@@ -5935,11 +5712,7 @@ end
 
 % Define plain chapter starts, and page on/off switching for it.
 \def\chapbreak{\dobreak \chapheadingskip {-4000}}
-
-% Start a new page
 \def\chappager{\par\vfill\supereject}
-
-% \chapoddpage - start on an odd page for a new chapter
 % Because \domark is called before \chapoddpage, the filler page will
 % get the headings for the next chapter, which is wrong.  But we don't
 % care -- we just disable all headings on the filler page.
@@ -5975,7 +5748,7 @@ end
 
 \CHAPPAGon
 
-% \chapmacro - Chapter opening.
+% Chapter opening.
 %
 % #1 is the text, #2 is the section type (Ynumbered, Ynothing,
 % Yappendix, Yomitfromtoc), #3 the chapter number.
@@ -6369,7 +6142,7 @@ end
   \savepageno = \pageno
   \begingroup                  % Set up to handle contents files properly.
     \raggedbottom              % Worry more about breakpoints than the bottom.
-    \entryrightmargin=\contentsrightmargin % Don't use the full line length.
+    \advance\hsize by -\contentsrightmargin % Don't use the full line length.
     %
     % Roman numerals for page numbers.
     \ifnum \pageno>0 \global\pageno = \lastnegativepageno \fi
@@ -6474,7 +6247,7 @@ end
 
 % Chapters, in the main contents.
 \def\numchapentry#1#2#3#4{\dochapentry{#2\labelspace#1}{#4}}
-
+%
 % Chapters, in the short toc.
 % See comments in \dochapentry re vbox and related settings.
 \def\shortchapentry#1#2#3#4{%
@@ -6489,7 +6262,7 @@ end
   \setbox0 = \hbox{\putwordAppendix{} M}%
   \hbox to \wd0{\putwordAppendix{} #1\hss}}
 %
-\def\appentry#1#2#3#4{\dochapentry{\appendixbox{#2}\hskip.7em#1}{#4}}
+\def\appentry#1#2#3#4{\dochapentry{\appendixbox{#2}\labelspace#1}{#4}}
 
 % Unnumbered chapters.
 \def\unnchapentry#1#2#3#4{\dochapentry{#1}{#4}}
@@ -6522,8 +6295,6 @@ end
 \def\dochapentry#1#2{%
    \penalty-300 \vskip1\baselineskip plus.33\baselineskip minus.25\baselineskip
    \begingroup
-     % Move the page numbers slightly to the right
-     \advance\entryrightmargin by -0.05em
      \chapentryfonts
      \tocentry{#1}{\dopageno\bgroup#2\egroup}%
    \endgroup
@@ -6643,24 +6414,6 @@ end
     \endgraf
     \ifdim\lastskip<\envskipamount
       \removelastskip
-      \ifnum\lastpenalty<10000
-        % Penalize breaking before the environment, because preceding text
-        % often leads into it.
-        \penalty100
-      \fi
-      \vskip\envskipamount
-    \fi
-  \fi
-}}
-
-\def\afterenvbreak{{%
-  % =10000 instead of <10000 because of a special case in \itemzzz and
-  % \sectionheading, q.v.
-  \ifnum \lastpenalty=10000 \else
-    \advance\envskipamount by \parskip
-    \endgraf
-    \ifdim\lastskip<\envskipamount
-      \removelastskip
       % it's not a good place to break if the last penalty was \nobreak
       % or better ...
       \ifnum\lastpenalty<10000 \penalty-50 \fi
@@ -6669,6 +6422,8 @@ end
   \fi
 }}
 
+\let\afterenvbreak = \aboveenvbreak
+
 % \nonarrowing is a flag.  If "set", @lisp etc don't narrow margins; it will
 % also clear it, so that its embedded environments do the narrowing again.
 \let\nonarrowing=\relax
@@ -6712,7 +6467,7 @@ end
   % collide with the section heading.
   \ifnum\lastpenalty>10000 \vskip\parskip \penalty\lastpenalty \fi
   %
-  \setbox\groupbox=\vbox\bgroup
+  \vbox\bgroup
       \baselineskip=0pt\parskip=0pt\lineskip=0pt
       \carttop
       \hbox\bgroup
@@ -6736,7 +6491,6 @@ end
       \egroup
       \cartbot
   \egroup
-  \addgroupbox
   \checkinserts
 }
 
@@ -7119,7 +6873,7 @@ end
 % typesetting commands (@smallbook, font changes, etc.) have to be done
 % beforehand -- and a) we want @copying to be done first in the source
 % file; b) letting users define the frontmatter in as flexible order as
-% possible is desirable.
+% possible is very desirable.
 %
 \def\copying{\checkenv{}\begingroup\scanargctxt\docopying}
 address@hidden copying{\endgroup\def\copyingtext{#1}}
@@ -7214,7 +6968,7 @@ end
   \temp
 }
 
-% \domakedefun \deffn \deffnx \deffnheader { (defn. of \deffnheader) }
+% \domakedefun \deffn \deffnx \deffnheader
 %
 % Define \deffn and \deffnx, without parameters.
 % \deffnheader has to be defined explicitly.
@@ -7552,44 +7306,34 @@ end
   }
 \fi
 
-\let\aftermacroxxx\relax
-\def\aftermacro{\aftermacroxxx}
-
-% alias because \c means cedilla in @tex or @math
-\let\texinfoc=\c
-
-% Used at the time of macro expansion.
-% Argument is macro body with arguments substituted
-\def\scanmacro#1{%
+\def\scanmacro#1{\begingroup
   \newlinechar`\^^M
   \let\xeatspaces\eatspaces
-  % Reduce doubled backslashes to one
-  \def\xprocessmacroarg{\passargtomacro\eatspaces}%
   %
-  % Process the macro body under the current catcode regime.
-  \scantokens{#1\texinfoc}\aftermacro%
+  % Undo catcode changes of \startcontents and \doprintindex
+  % When called from @insertcopying or (short)caption, we need active
+  % backslash to get it printed correctly.  Previously, we had
+  % \catcode`\\=\other instead.  We'll see whether a problem appears
+  % with macro expansion.                              --kasal, 19aug04
+  address@hidden \catcode`\\=\active \escapechar=`\@
   %
-  % The \c is to remove the \newlinechar added by \scantokens, and
-  % can be noticed by \parsearg.
-  %   The \aftermacro allows a \comment at the end of the macro definition
-  % to duplicate itself past the final \newlinechar added by \scantokens:
-  % this is used in the definition of \group to comment out a newline.  We
-  % don't do the same for \c to support Texinfo files with macros that ended
-  % with a @c, which should no longer be necessary.
-  %   We avoid surrounding the call to \scantokens with \bgroup and \egroup
-  % to allow macros to open or close groups themselves.
-}
+  % ... and for \example:
+  \spaceisspace
+  %
+  % The \empty here causes a following catcode 5 newline to be eaten as
+  % part of reading whitespace after a control sequence.  It does not
+  % eat a catcode 13 newline.  There's no good way to handle the two
+  % cases (untried: maybe e-TeX's \everyeof could help, though plain TeX
+  % would then have different behavior).  See the Macro Details node in
+  % the manual for the workaround we recommend for macros and
+  % line-oriented commands.
+  % 
+  \scantokens{#1\empty}%
+\endgroup}
 
 \def\scanexp#1{%
-  \bgroup
-  % Undo catcode changes of \startcontents and \printindex
-  % When called from @insertcopying or (short)caption, we need active
-  % backslash to get it printed correctly.
-  % FIXME: This may not be needed.
-  address@hidden \catcode`\\=\active \escapechar=`\@
   \edef\temp{\noexpand\scanmacro{#1}}%
   \temp
-  \egroup
 }
 
 \newcount\paramno   % Count of parameters
@@ -7655,6 +7399,7 @@ end
   \catcode`\+=\other
   \catcode`\<=\other
   \catcode`\>=\other
+  address@hidden
   \catcode`\^=\other
   \catcode`\_=\other
   \catcode`\|=\other
@@ -7664,35 +7409,37 @@ end
 
 \def\scanargctxt{% used for copying and captions, not macros.
   \scanctxt
-  address@hidden
   \catcode`\\=\other
   \catcode`\^^M=\other
 }
 
 \def\macrobodyctxt{% used for @macro definitions
   \scanctxt
-  \catcode`\ =\other
-  address@hidden
   \catcode`\{=\other
   \catcode`\}=\other
   \catcode`\^^M=\other
   \usembodybackslash
 }
 
-% Used when scanning braced macro arguments.  Note, however, that catcode
-% changes here are ineffectual if the macro invocation was nested inside
-% an argument to another Texinfo command.
-\def\macroargctxt{%
-  \scanctxt
-  \catcode`\^^M=\other
-  \catcode`\\=\active
-}
-
-\def\macrolineargctxt{% used for whole-line arguments without braces
+\def\macroargctxt{% used when scanning invocations
   \scanctxt
-  \catcode`\{=\other
-  \catcode`\}=\other
+  \catcode`\\=0
 }
+% why catcode 0 for \ in the above?  To recognize \\ \{ \} as "escapes"
+% for the single characters \ { }.  Thus, we end up with the "commands"
+% that would be written @\ @{ @} in a Texinfo document.
+% 
+% We already have @{ and @}.  For @\, we define it here, and only for
+% this purpose, to produce a typewriter backslash (so, the @\ that we
+% define for @math can't be used with @macro calls):
+%
+\def\\{\normalbackslash}%
+% 
+% We would like to do this for \, too, since that is what makeinfo does.
+% But it is not possible, because Texinfo already has a command @, for a
+% cedilla accent.  Documents must use @comma{} instead.
+%
+% \anythingelse will almost certainly be an error of some kind.
 
 % \mbodybackslash is the definition of \ in @macro bodies.
 % It maps \foo\ => \csname macarg.foo\endcsname => #N
@@ -7764,36 +7511,56 @@ end
   \fi
 }
 
-% \getargs -- Parse the arguments to a @macro line.  Set \macname to
-% the name of the macro, and \argl to the braced argument list.
+% This makes use of the obscure feature that if the last token of a
+% <parameter list> is #, then the preceding argument is delimited by
+% an opening brace, and that opening brace is not consumed.
 \def\getargs#1{\getargsxxx#1{}}
 \def\getargsxxx#1#{\getmacname #1 \relax\getmacargs}
 \def\getmacname#1 #2\relax{\macname={#1}}
 \def\getmacargs#1{\def\argl{#1}}
-% This made use of the feature that if the last token of a
-% <parameter list> is #, then the preceding argument is delimited by
-% an opening brace, and that opening brace is not consumed.
 
-% Parse the optional {params} list to @macro or @rmacro.
-% Set \paramno to the number of arguments,
-% and \paramlist to a parameter text for the macro (e.g. #1,#2,#3 for a
-% three-param macro.)  Define \macarg.BLAH for each BLAH in the params
-% list to some hook where the argument is to be expanded.  If there are
-% less than 10 arguments that hook is to be replaced by ##N where N
+% For macro processing make @ a letter so that we can make Texinfo private 
macro names.
address@hidden
+\catcode address@hidden
+
+% Parse the optional {params} list.  Set up \paramno and \paramlist
+% so \defmacro knows what to do.  Define \macarg.BLAH for each BLAH
+% in the params list to some hook where the argument is to be expanded.  If
+% there are less than 10 arguments that hook is to be replaced by ##N where N
 % is the position in that list, that is to say the macro arguments are to be
 % defined `a la TeX in the macro body.  
 %
 % That gets used by \mbodybackslash (above).
 %
-% If there are 10 or more arguments, a different technique is used: see
-% \parsemmanyargdef.
+% We need to get `macro parameter char #' into several definitions.
+% The technique used is stolen from LaTeX: let \hash be something
+% unexpandable, insert that wherever you need a #, and then redefine
+% it to # just before using the token list produced.
 %
+% The same technique is used to protect \eatspaces till just before
+% the macro is used.
+%
+% If there are 10 or more arguments, a different technique is used, where the
+% hook remains in the body, and when macro is to be expanded the body is
+% processed again to replace the arguments.
+%
+% In that case, the hook is \the\toks N-1, and we simply set \toks N-1 to the
+% argument N value and then \edef  the body (nothing else will expand because 
of
+% the catcode regime underwhich the body was input).
+%
+% If you compile with TeX (not eTeX), and you have macros with 10 or more
+% arguments, no macro can have more than 256 arguments (else error).
 \def\parsemargdef#1;{%
   \paramno=0\def\paramlist{}%
   \let\hash\relax
-  % \hash is redefined to `#' later to get it into definitions
-  \let\processmacroarg\relax
+  \let\xeatspaces\relax
   \parsemargdefxxx#1,;,%
+  % In case that there are 10 or more arguments we parse again the arguments
+  % list to set new definitions for the \macarg.BLAH macros corresponding to
+  % each BLAH argument. It was anyhow needed to parse already once this list
+  % in order to count the arguments, and as macros with at most 9 arguments
+  % are by far more frequent than macro with 10 or more arguments, defining
+  % twice the \macarg.BLAH macros does not cost too much processing power.
   \ifnum\paramno<10\relax\else
     \paramno0\relax
     \parsemmanyargdef@@#1,;,% 10 or more arguments
@@ -7804,47 +7571,10 @@ end
   \else \let\next=\parsemargdefxxx
     \advance\paramno by 1
     \expandafter\edef\csname macarg.\eatspaces{#1}\endcsname
-        {\processmacroarg{\hash\the\paramno}}%
+        {\xeatspaces{\hash\the\paramno}}%
     \edef\paramlist{\paramlist\hash\the\paramno,}%
   \fi\next}
 
-% \parsemacbody, \parsermacbody
-%
-% Read recursive and nonrecursive macro bodies. (They're different since
-% rec and nonrec macros end differently.)
-% 
-% We are in \macrobodyctxt, and the \xdef causes backslashshes in the macro 
-% body to be transformed.
-% Set \macrobody to the body of the macro, and call \defmacro.
-%
-{\catcode`\ address@hidden macro{%
-\xdef\macrobody{\eatcr{#1}}\endgroup\defmacro}}%
-{\catcode`\ address@hidden rmacro{%
-\xdef\macrobody{\eatcr{#1}}\endgroup\defmacro}}%
-
-% Make @ a letter, so that we can make private-to-Texinfo macro names.
address@hidden
-\catcode address@hidden
-
-%%%%%%%%%%%%%% Code for > 10 arguments only   %%%%%%%%%%%%%%%%%%
-
-% If there are 10 or more arguments, a different technique is used, where the
-% hook remains in the body, and when macro is to be expanded the body is
-% processed again to replace the arguments.
-%
-% In that case, the hook is \the\toks N-1, and we simply set \toks N-1 to the
-% argument N value and then \edef the body (nothing else will expand because of
-% the catcode regime under which the body was input).
-%
-% If you compile with TeX (not eTeX), and you have macros with 10 or more
-% arguments, no macro can have more than 256 arguments (else error).
-%
-% In case that there are 10 or more arguments we parse again the arguments
-% list to set new definitions for the \macarg.BLAH macros corresponding to
-% each BLAH argument. It was anyhow needed to parse already once this list
-% in order to count the arguments, and as macros with at most 9 arguments
-% are by far more frequent than macro with 10 or more arguments, defining
-% twice the \macarg.BLAH macros does not cost too much processing power.
 \def\parsemmanyargdef@@#1,{%
   \if#1;\let\next=\relax
   \else 
@@ -7860,6 +7590,16 @@ end
     \advance\paramno by 1\relax
   \fi\next}
 
+% These two commands read recursive and nonrecursive macro bodies.
+% (They're different since rec and nonrec macros end differently.)
+%
+
+\catcode address@hidden
address@hidden macro%
+{\xdef\temp{\eatcr{#1}}\endgroup\defmacro}%
address@hidden rmacro%
+{\xdef\temp{\eatcr{#1}}\endgroup\defmacro}%
+\catcode address@hidden
 
 address@hidden
 address@hidden
@@ -7867,7 +7607,7 @@ end
 address@hidden@}%
 
 % This macro is expanded during the Texinfo macro expansion, not during its
-% definition.  It gets all the arguments' values and assigns them to macros
+% definition.  It gets all the arguments values and assigns them to macros
 % macarg.ARGNAME
 %
 % #1 is the macro name
@@ -7888,6 +7628,8 @@ end
     \getargvals@@
   \fi
 }
+
+% 
 \def\getargvals@@{%
   \ifx\paramlist\nilm@
       % Some sanity check needed here that \argvaluelist is also empty.
@@ -7931,8 +7673,7 @@ end
 }
 
 % Replace arguments by their values in the macro body, and place the result
-% in macro address@hidden
-% 
+% in macro address@hidden
 address@hidden
   %  To do this we use the property that token registers that are \the'ed
   % within an \edef  expand only once. So we are going to place all argument
@@ -7956,9 +7697,8 @@ end
   address@hidden@tempc}%
   }
 
-% Define the named-macro outside of this group and then close this group. 
-% 
 address@hidden 
+  %% Define the named-macro outside of this group and then close this group. 
   \expandafter
   \endgroup
   \macargdeflist@
@@ -7995,8 +7735,14 @@ end
   \next
 }
 
-% Trailing missing arguments are set to empty.
-% 
+% Save the token stack pointer into macro #1
address@hidden
+% Restore the token stack pointer from number in macro #1
address@hidden
+% newtoks that can be used non \outer .
+\def\texinonouternewtoks{\alloc@ 5\toks \toksdef address@hidden
+
+% Tailing missing arguments are set to empty
 address@hidden
   \ifx\paramlist\nilm@
     \let\next\macargexpandinbody@
@@ -8026,204 +7772,99 @@ end
    \long\def#2{#4}%
 }
 
-
-%%%%%%%%%%%%%% End of code for > 10 arguments %%%%%%%%%%%%%%%%%%
-
-
-
-% Remove following spaces at the expansion stage.
-% This works because spaces are discarded before each argument when TeX is 
-% getting the arguments for a macro.
-% This must not be immediately followed by a }.
-\long\def\gobblespaces#1{#1}
-
-% This defines a Texinfo @macro or @rmacro, called by \parsemacbody.
-%    \macrobody has the body of the macro in it, with placeholders for
-% its parameters, looking like "\processmacroarg{\hash 1}".
-%    \paramno is the number of parameters
-%    \paramlist is a TeX parameter text, e.g. "#1,#2,#3,"
-% There are eight cases: recursive and nonrecursive macros of zero, one,
-% up to nine, and many arguments.
+% This defines a Texinfo @macro. There are eight cases: recursive and
+% nonrecursive macros of zero, one, up to nine, and many arguments.
+% Much magic with \expandafter here.
 % \xdef is used so that macro definitions will survive the file
-% they're defined in: @include reads the file inside a group.
+% they're defined in; @include reads the file inside a group.
 %
 \def\defmacro{%
   \let\hash=##% convert placeholders to macro parameter chars
-  \ifnum\paramno=1
-    \def\processmacroarg{\gobblespaces}%
-    % This removes the pair of braces around the argument.  We don't
-    % use \eatspaces, because this can cause ends of lines to be lost
-    % when the argument to \eatspaces is read, leading to line-based
-    % commands like "@itemize" not being read correctly.
-  \else
-    \def\processmacroarg{\xprocessmacroarg}%
-    \let\xprocessmacroarg\relax
-  \fi
-  \ifrecursive   %%%%%%%%%%%%%% Recursive %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+  \ifrecursive
     \ifcase\paramno
     % 0
       \expandafter\xdef\csname\the\macname\endcsname{%
-        \noexpand\scanmacro{\macrobody}}%
+        \noexpand\scanmacro{\temp}}%
     \or % 1
       \expandafter\xdef\csname\the\macname\endcsname{%
-         \bgroup
+         \bgroup\noexpand\macroargctxt
          \noexpand\braceorline
-         \expandafter\noexpand\csname\the\macname @@@\endcsname}%
-      \expandafter\xdef\csname\the\macname @@@\endcsname##1{%
-         \expandafter\noexpand\csname\the\macname @@@@\endcsname{%
-           \noexpand\gobblespaces##1\empty}%
-           % The \empty is for \gobblespaces in case #1 is empty
-         }%
-      \expandafter\xdef\csname\the\macname @@@@\endcsname##1{%
-         \egroup\noexpand\scanmacro{\macrobody}}%
+         \expandafter\noexpand\csname\the\macname xxx\endcsname}%
+      \expandafter\xdef\csname\the\macname xxx\endcsname##1{%
+         \egroup\noexpand\scanmacro{\temp}}%
     \else
       \ifnum\paramno<10\relax % at most 9
         \expandafter\xdef\csname\the\macname\endcsname{%
            \bgroup\noexpand\macroargctxt
-           \noexpand\csname\the\macname @@\endcsname}%
-        \expandafter\xdef\csname\the\macname @@\endcsname##1{%
-            \expandafter\noexpand\csname\the\macname @@@\endcsname ##1,}%
+           \noexpand\csname\the\macname xx\endcsname}%
+        \expandafter\xdef\csname\the\macname xx\endcsname##1{%
+            \expandafter\noexpand\csname\the\macname xxx\endcsname ##1,}%
         \expandafter\expandafter
         \expandafter\xdef
         \expandafter\expandafter
-          \csname\the\macname @@@\endcsname
-            \paramlist{\egroup\noexpand\scanmacro{\macrobody}}%
+          \csname\the\macname xxx\endcsname
+            \paramlist{\egroup\noexpand\scanmacro{\temp}}%
       \else % 10 or more
         \expandafter\xdef\csname\the\macname\endcsname{%
           address@hidden
         }%    
-        \global\expandafter\let\csname mac.\the\macname 
.body\endcsname\macrobody
+        \global\expandafter\let\csname mac.\the\macname .body\endcsname\temp
         \global\expandafter\let\csname mac.\the\macname 
.recurse\endcsname\gobble
       \fi
     \fi
-  \else  %%%%%%%%%%%%%%%%%%%%%% Non-recursive %%%%%%%%%%%%%%%%%%%%%%%%%%
+  \else
     \ifcase\paramno
     % 0
       \expandafter\xdef\csname\the\macname\endcsname{%
-        \noexpand\scanmacro{\macrobody}}%
+        \noexpand\norecurse{\the\macname}%
+        \noexpand\scanmacro{\temp}\egroup}%
     \or % 1
       \expandafter\xdef\csname\the\macname\endcsname{%
-         \bgroup
+         \bgroup\noexpand\macroargctxt
          \noexpand\braceorline
-         \expandafter\noexpand\csname\the\macname @@@\endcsname}%
-      \expandafter\xdef\csname\the\macname @@@\endcsname##1{%
-         \expandafter\noexpand\csname\the\macname @@@@\endcsname{%
-           \noexpand\gobblespaces##1\empty}%
-           % The \empty is for \gobblespaces in case #1 is empty
-         }%
-      \expandafter\xdef\csname\the\macname @@@@\endcsname##1{%
+         \expandafter\noexpand\csname\the\macname xxx\endcsname}%
+      \expandafter\xdef\csname\the\macname xxx\endcsname##1{%
         \egroup
-        \noexpand\scanmacro{\macrobody}%
-        }%
+        \noexpand\norecurse{\the\macname}%
+        \noexpand\scanmacro{\temp}\egroup}%
     \else % at most 9
       \ifnum\paramno<10\relax
         \expandafter\xdef\csname\the\macname\endcsname{%
            \bgroup\noexpand\macroargctxt
-           \expandafter\noexpand\csname\the\macname @@\endcsname}%
-        \expandafter\xdef\csname\the\macname @@\endcsname##1{%
-            \expandafter\noexpand\csname\the\macname @@@\endcsname ##1,}%
+           \expandafter\noexpand\csname\the\macname xx\endcsname}%
+        \expandafter\xdef\csname\the\macname xx\endcsname##1{%
+            \expandafter\noexpand\csname\the\macname xxx\endcsname ##1,}%
         \expandafter\expandafter
         \expandafter\xdef
         \expandafter\expandafter
-        \csname\the\macname @@@\endcsname
+        \csname\the\macname xxx\endcsname
         \paramlist{%
             \egroup
-            \noexpand\scanmacro{\macrobody}%
-            }%
+            \noexpand\norecurse{\the\macname}%
+            \noexpand\scanmacro{\temp}\egroup}%
       \else % 10 or more:
         \expandafter\xdef\csname\the\macname\endcsname{%
           address@hidden
         }%
-        \global\expandafter\let\csname mac.\the\macname 
.body\endcsname\macrobody
+        \global\expandafter\let\csname mac.\the\macname .body\endcsname\temp
         \global\expandafter\let\csname mac.\the\macname 
.recurse\endcsname\norecurse
       \fi
     \fi
   \fi}
 
-\catcode address@hidden % end private-to-Texinfo catcodes
+\catcode address@hidden
 
 \def\norecurse#1{\bgroup\cslet{#1}{macsave.#1}}
 
-
address@hidden \catcode`\\=13
address@hidden@_=11
-
-% Call #1 with a list of tokens #2, with any doubled backslashes in #2
-% compressed to one.
address@hidden@passargtomacro#1#2{%
-  @address@hidden
-  @address@hidden
-  @address@hidden@finish}%
-  @address@hidden
-  @address@hidden@relax
-  @address@hidden
-}
-
-% Input stream is just after a backslash.  If the next token is not a
-% backslash, process the rest of the argument; otherwise, remove the next
-% token.
address@hidden@look_ahead{%
-  @address@hidden@look_aheadzzz}
address@hidden@look_aheadzzz{%
-  @address@hidden
-   @address@hidden@gobble_and_check_finish 
-  @else
-   @address@hidden@add_segment
-  @address@hidden
-}
-
-% Double backslash found.  Add a single backslash here.
address@hidden@gobble_and_check_finish#1{%
-  @add_the_backslash
-  @address@hidden
-  @address@hidden@add_segment
-}
-
-% append a backslash to \arg_result
address@hidden@add_the_backslash{%
-  @address@hidden@address@hidden@address@hidden
-}
-
-% Input stream is either at the start of the argument, or just after a 
-% backslash sequence, either a lone backslash, or a doubled backslash.  
-% \next_token contains the first token in the input stream: if it is \finish, 
-% finish; otherwise, append to \arg_result the segment of the argument up until
-% the next backslash.  \pending_backslash contains a backslash to represent
-% a backslash just before the start of the input stream that has not been
-% added to \arg_result.
address@hidden@add_segment#1\{%
address@hidden@address@hidden
-  @address@hidden@call_the_macro%
address@hidden
-  @address@hidden@look_ahead
-  %
-  % append to @arg_result
-  % token list registers might be better
-  @address@hidden@address@hidden
-  @address@hidden@address@hidden
-  @address@hidden@expandafter{%
-  @address@hidden
-  @pending_backslash#1}%
-  @address@hidden
address@hidden@next}
-
address@hidden@address@hidden@address@hidden@arg_result}}
-
-}
-
-% \braceorline MAC is used for a one-argument macro MAC.  It checks
-% whether the next non-whitespace character is a {.  It sets the context
-% for reading the argument (slightly different in the two cases).  Then,
-% to read the argument, in the whole-line case, it then calls the regular
-% \parsearg MAC; in the lbrace case, it calls \passargtomacro MAC.
+% \braceorline decides whether the next nonwhitespace character is a
+% {.  If so it reads up to the closing }, if not, it reads the whole
+% line.  Whatever was read is then fed to the next control sequence
+% as an argument (by \parsebrace or \parsearg).
 % 
 \def\braceorline#1{\let\macnamexxx=#1\futurelet\nchar\braceorlinexxx}
 \def\braceorlinexxx{%
-  \ifx\nchar\bgroup
-    \macroargctxt
-    \expandafter\passargtomacro
-  \else
-    \macrolineargctxt\expandafter\parsearg
+  \ifx\nchar\bgroup\else
+    \expandafter\parsearg
   \fi \macnamexxx}
 
 
@@ -8305,7 +7946,6 @@ end
   \pdfmkdest{#1}%
   \iflinks
     {%
-      \requireauxfile
       \atdummies  % preserve commands, but don't expand them
       \edef\writexrdef##1##2{%
        address@hidden #1 of \setref, expanded by the \edef
@@ -8345,12 +7985,9 @@ end
 % node name, #4 the name of the Info file, #5 the name of the printed
 % manual.  All but the node name can be omitted.
 %
-\def\pxref{\putwordsee{} \xrefXX}
-\def\xref{\putwordSee{} \xrefXX}
-\def\ref{\xrefXX}
-
-\def\xrefXX#1{\def\xrefXXarg{#1}\futurelet\tokenafterxref\xrefXXX}
-\def\xrefXXX{\expandafter\xrefX\expandafter[\xrefXXarg,,,,,,,]}
+\def\pxref#1{\putwordsee{} \xrefX[#1,,,,,,,]}
+\def\xref#1{\putwordSee{} \xrefX[#1,,,,,,,]}
+\def\ref#1{\xrefX[#1,,,,,,,]}
 %
 \newbox\toprefbox
 \newbox\printedrefnamebox
@@ -8494,12 +8131,6 @@ end
       %
       % output the `page 3'.
       \turnoffactive \putwordpage\tie\refx{#1-pg}{}%
-           \ifx,\tokenafterxref
-      \else\ifx.\tokenafterxref
-      \else\ifx;\tokenafterxref
-      \else\ifx)\tokenafterxref
-      \else,% add a , if xref not followed by punctuation
-      \fi\fi\fi\fi
     \fi\fi
   \fi
   \endlink
@@ -8570,7 +8201,6 @@ end
 % If its value is nonempty, SUFFIX is output afterward.
 %
 \def\refx#1#2{%
-  \requireauxfile
   {%
     \indexnofonts
     \otherbackslash
@@ -8634,23 +8264,6 @@ end
   \fi
 }
 
-% If working on a large document in chapters, it is convenient to
-% be able to disable indexing, cross-referencing, and contents, for test runs.
-% This is done with @novalidate at the beginning of the file.
-%
-\newif\iflinks \linkstrue % by default we want the aux files.
-\let\novalidate = \linksfalse
-
-% Used when writing to the aux file, or when using data from it.
-\def\requireauxfile{%
-  \iflinks
-    \tryauxfile
-    % Open the new aux file.  TeX will close it automatically at exit.
-    \immediate\openout\auxfile=\jobname.aux
-  \fi
-  \global\let\requireauxfile=\relax   % Only do this once.
-}
-
 % Read the last existing aux file, if any.  No error if none exists.
 %
 \def\tryauxfile{%
@@ -9124,7 +8737,6 @@ end
       % \floatlabel-lof.  Besides \floatident, we include the short
       % caption if specified, else the full caption if specified, else nothing.
       {%
-        \requireauxfile
         \atdummies
         %
         % since we read the caption text in the macro world, where ^^M
@@ -9342,6 +8954,7 @@ directory should work if nowhere else does.}
       \catcode\count255=#1\relax
       \advance\count255 by 1
    \repeat
+
 }
 
 % @documentencoding sets the definition of non-ASCII characters
@@ -9408,17 +9021,17 @@ directory should work if nowhere else does.}
 \def\latonechardefs{%
   \gdef^^a0{\tie}
   \gdef^^a1{\exclamdown}
-  \gdef^^a2{{\tcfont \char162}} % cent
-  \gdef^^a3{\pounds}
-  \gdef^^a4{{\tcfont \char164}} % currency
-  \gdef^^a5{{\tcfont \char165}} % yen
-  \gdef^^a6{{\tcfont \char166}} % broken bar
+  \gdef^^a2{\missingcharmsg{CENT SIGN}}
+  \gdef^^a3{{\pounds}}
+  \gdef^^a4{\missingcharmsg{CURRENCY SIGN}}
+  \gdef^^a5{\missingcharmsg{YEN SIGN}}
+  \gdef^^a6{\missingcharmsg{BROKEN BAR}}
   \gdef^^a7{\S}
   \gdef^^a8{\"{}}
   \gdef^^a9{\copyright}
   \gdef^^aa{\ordf}
   \gdef^^ab{\guillemetleft}
-  \gdef^^ac{\ensuremath\lnot}
+  \gdef^^ac{$\lnot$}
   \gdef^^ad{\-}
   \gdef^^ae{\registeredsymbol}
   \gdef^^af{\={}}
@@ -9783,11 +9396,7 @@ directory should work if nowhere else does.}
 \def\utfeightchardefs{%
   \DeclareUnicodeCharacter{00A0}{\tie}
   \DeclareUnicodeCharacter{00A1}{\exclamdown}
-  \DeclareUnicodeCharacter{00A2}{{\tcfont \char162}}% 0242=cent
   \DeclareUnicodeCharacter{00A3}{\pounds}
-  \DeclareUnicodeCharacter{00A4}{{\tcfont \char164}}% 0244=currency
-  \DeclareUnicodeCharacter{00A5}{{\tcfont \char165}}% 0245=yen
-  \DeclareUnicodeCharacter{00A6}{{\tcfont \char166}}% 0246=brokenbar
   \DeclareUnicodeCharacter{00A7}{\S}
   \DeclareUnicodeCharacter{00A8}{\"{ }}
   \DeclareUnicodeCharacter{00A9}{\copyright}
@@ -10550,8 +10159,8 @@ directory should work if nowhere else does.}
 \def\activedoublequote{{\tt\char34}}
 \let"=\activedoublequote
 \catcode`\~=\active \def\activetilde{{\tt\char126}} \let~ = \activetilde
-\chardef\hatchar=`\^
-\catcode`\^=\active \def\activehat{{\tt \hatchar}} \let^ = \activehat
+\chardef\hat=`\^
+\catcode`\^=\active \def\activehat{{\tt \hat}} \let^ = \activehat
 
 \catcode`\_=\active
 \def_{\ifusingtt\normalunderscore\_}
@@ -10568,8 +10177,6 @@ directory should work if nowhere else does.}
 \catcode`\>=\active \def\activegtr{{\tt \gtr}}\let> = \activegtr
 \catcode`\+=\active \def+{{\tt \char 43}}
 \catcode`\$=\active \def${\ifusingit{{\sl\$}}\normaldollar}%$ font-lock fix
-\catcode`\-=\active \let-=\normaldash
-
 
 % used for headline/footline in the output routine, in case the page
 % breaks in the middle of an @tex block.
@@ -10584,6 +10191,12 @@ directory should work if nowhere else does.}
   % in principle, all other definitions in \tex have to be undone too.
 }
 
+% If a .fmt file is being used, characters that might appear in a file
+% name cannot be active until we have parsed the command line.
+% So turn them off again, and have \everyjob (or @setfilename) turn them on.
+% \otherifyactive is called near the end of this file.
+\def\otherifyactive{\catcode`+=\other \catcode`\_=\other}
+
 % Used sometimes to turn off (effectively) the active characters even after
 % parsing them.
 \def\turnoffactive{%
@@ -10602,22 +10215,23 @@ directory should work if nowhere else does.}
 % \doublebackslash is two of them (for the pdf outlines).
 {\catcode`\\=\other @address@hidden @address@hidden
 
-% In Texinfo, backslash is an active character; it prints the backslash
+% In texinfo, backslash is an active character; it prints the backslash
 % in fixed width font.
 \catcode`\\=\active  % @ for escape char from now on.
 
-% Print a typewriter backslash.  For math mode, we can't simply use
-% \backslashcurfont: the story here is that in math mode, the \char
-% of \backslashcurfont ends up printing the roman \ from the math symbol
-% font (because \char in math mode uses the \mathcode, and plain.tex
-% sets \mathcode`\\="026E).  Hence we use an explicit \mathchar,
+% The story here is that in math mode, the \char of \backslashcurfont
+% ends up printing the roman \ from the math symbol font (because \char
+% in math mode uses the \mathcode, and plain.tex sets
+% \mathcode`\\="026E).  It seems better for @backslashchar{} to always
+% print a typewriter backslash, hence we use an explicit \mathchar,
 % which is the decimal equivalent of "715c (class 7, e.g., use \fam;
 % ignored family value; char position "5C).  We can't use " for the
 % usual hex value because it has already been made active.
address@hidden@address@hidden @ifmmode @mathchar29020 @else @backslashcurfont 
@fi}}
address@hidden@backslashchar = @normalbackslash % @backslashchar{} is for user 
documents.
 
address@hidden@address@hidden @ifmmode @mathchar29020 @else @backslashcurfont 
@fi}}
address@hidden@backslashchar = @ttbackslash % @backslashchar{} is for user 
documents.
-
+% On startup, @fixbackslash assigns:
+%  @let \ = @normalbackslash
 % \rawbackslash defines an active \ to do \backslashcurfont.
 % \otherbackslash defines an active \ to be a literal `\' character with
 % catcode other.  We switch back and forth between these.
@@ -10625,7 +10239,8 @@ directory should work if nowhere else does.}
 @address@hidden@address@hidden
 
 % Same as @turnoffactive except outputs \ as {\tt\char`\\} instead of
-% the literal character `\'.
+% the literal character `\'.  Also revert - to its normal character, in
+% case the active - from code has slipped in.
 %
 address@hidden = @active
  @address@hidden
@@ -10636,66 +10251,40 @@ directory should work if nowhere else does.}
    @address@hidden
    @let<address@hidden
    @let>address@hidden
+   @address@hidden
    @address@hidden
    @address@hidden
    @let|address@hidden
    @address@hidden
-   @address@hidden
    @markupsetuplqdefault
    @markupsetuprqdefault
    @unsepspaces
  }
 }
 
-% If a .fmt file is being used, characters that might appear in a file
-% name cannot be active until we have parsed the command line.
-% So turn them off again, and have @fixbackslash turn them back on.
address@hidden@other @address@hidden@other
+% Make _ and + \other characters, temporarily.
+% This is canceled by @fixbackslash.
address@hidden
 
-% \enablebackslashhack - allow file to begin `\input texinfo'
-%
 % If a .fmt file is being used, we don't want the `\input texinfo' to show up.
 % That is what \eatinput is for; after that, the `\' should revert to printing
 % a backslash.
-% If the file did not have a `\input texinfo', then it is turned off after
-% the first line; otherwise the first `\' in the file would cause an error.
-% This is used on the very last line of this file, texinfo.tex.
-% We also use @c to call @fixbackslash, in case ends of lines are hidden.
-{
address@hidden@^=7
address@hidden@address@hidden@enablebackslashhack{%
-  @address@hidden = @eatinput%
-  @address@hidden
-  @address@hidden@address@hidden
-  @def address@hidden@secondlinenl}%
-  @gdef @address@hidden@thirdlinenl}%
-  @gdef @address@hidden
-}}
-
address@hidden@^=7 @address@hidden
address@hidden@eatinput input address@hidden
+%
address@hidden@eatinput input address@hidden
address@hidden@let\ = @eatinput
 
+% On the other hand, perhaps the file did not have a `\input texinfo'. Then
+% the first `\' in the file would cause an error. This macro tries to fix
+% that, assuming it is called before the first `\' could plausibly occur.
+% Also turn back on active characters that might appear in the input
+% file name, in case not using a pre-dumped format.
+%
 @address@hidden
-  @address@hidden @let\ = @ttbackslash @fi
-  @catcode13=5 % regular end of line
-  @address@hidden@texinfoc
-  % Also turn back on active characters that might appear in the input
-  % file name, in case not using a pre-dumped format.
+  @address@hidden @let\ = @normalbackslash @fi
   @address@hidden
   @address@hidden@active
-  %
-  % If texinfo.cnf is present on the system, read it.
-  % Useful for site-wide @afourpaper, etc.  This macro, @fixbackslash, gets
-  % called at the beginning of every Texinfo file.  Not opening texinfo.cnf
-  % directly in this file, texinfo.tex, makes it possible to make a format
-  % file for Texinfo.
-  %
-  @openin 1 texinfo.cnf
-  @ifeof 1 @else @input texinfo.cnf @fi
-  @closein 1
 }
 
-
 % Say @foo, not \foo, in error messages.
 @escapechar = `@@
 
@@ -10735,4 +10324,3 @@ directory should work if nowhere else does.}
 @ignore
    arch-tag: e1b36e32-c96e-4135-a41a-0b2efa2ea115
 @end ignore
address@hidden

-----------------------------------------------------------------------

Summary of changes:
 doc/ChangeLog   |    6 +
 doc/texinfo.tex | 1262 +++++++++++++++++++------------------------------------
 2 files changed, 431 insertions(+), 837 deletions(-)


hooks/post-receive
-- 
gawk



reply via email to

[Prev in Thread] Current Thread [Next in Thread]