Re: improved TeX output

[Werner LEMBERG]

> Next time, can you make sure that the manual actually builds?

Oops!  Sorry.  Here a patch for the patch which slipped through.

Note that due to the new behaviour texinfo can much better manipulate
LilyPond output.  Perhaps surprisingly, this produce worse results
currently (mostly caused by too much indentation).  I'll fix them in
the next days.


    Werner


======================================================================

--- lilyponddefs.tex.old        Sun Sep 15 21:45:14 2002
+++ lilyponddefs.tex    Mon Sep 16 09:32:08 2002
@@ -26,7 +26,9 @@
 
 % skip if included already
 
-\def\SkipLilydefs{\endinput}
+\def\SkipLilydefs{
+  \endlinechar \lilyponddefsELC
+  \endinput}
 \ifundefined{EndLilyPondOutput}
   \def\EndLilyPondOutput{\csname bye\endcsname}
   \def\SkipLilydefs{}
@@ -42,7 +44,7 @@
 \fi
 
 % The feta characters
-\input feta20.tex
+\input feta20
 
 \font\fetasixteen = feta16
 \def\fetafont{\fetasixteen}
@@ -55,8 +57,6 @@
   \hbox to 0pt{#1\hss}
 }
 
-\newdimen\scoreshift
-
 % Attempt to keep lilypiecetitle together with the piece:
 
 %
@@ -118,6 +118,22 @@
 \ifx\outputscale\undefined
   \csname global\endcsname\csname newdimen\endcsname\outputscale
 \fi
+
+\ifundefined{scoreshift}
+\else
+  % It is very ugly to hide \newdimen with \endinput, but I see no
+  % alternative: Since it is defined as \outer in plain.tex, you can
+  % neither use it in macros nor skip in an \if... \fi construction.
+  %
+  % In general, it is very bad that lilyponddefs.tex is read in again
+  % and again...  This should be fixed by putting the variable parts in
+  % this file into a macro so that loading the file multiple times can be
+  % avoided.
+  \endlinechar \lilyponddefsELC
+  \expandafter\endinput
+\fi
+
+\newdimen\scoreshift
 
 \endlinechar \lilyponddefsELC
 \endinput