We don't need major_release_split any more.

src/tools/RELEASE_CHANGES

@@ -68,15 +68,14 @@ Creating Back-Branch Release Notes
 
 * Do 'cvs log' on each back-branch and run pgcvslog
 
-* Edit and create SGML markup for the most recent branch
+* In CVS HEAD, edit and create SGML markup for the most recent branch
+  in that branch's release-N.N.sgml file
 
-* Make copies of the SGML for each branch, then remove items
-  that do not apply based on cvs logs for that branch
+* Copy this into older branches' release-N.N.sgml files, then remove
+  items that do not apply based on cvs logs for that branch (and add
+  any that are needed)
 
-* Copy the SGML for each branch into release.sgml in CVS HEAD
-
-* Use src/tools/major_release_split to create a release.sgml for
-  each branch and copy them to the branch's CVS
+* Copy the appropriate release-N.N.sgml files into each back CVS branch
 
 
 ---------------------------------------------------------------------------

src/tools/major_release_split

-:
-
-# This program takes release.sgml and breaks it up into
-# per-major-release files that can be copied to the proper
-# CVS tree.
-
-[ "$#" -ne 1 ] && echo "Usage: $0 release_sgml_file" 1>&2 && exit 1
-
-FILE="$1"
-
-trap "rm -f /tmp/preamble" 0 1 2 3 15
-
-# Create the SGML preamble file
-# Copy from the start of the file to the first "sect1" heading
-grep -B 1000000 "`sed -n '/<sect1/p;/<sect1/q' \"$FILE\"`" "$FILE" | 
-# exclude last line
-sed -n '$q;p' > /tmp/preamble
-
-# Create per-major-release files
-# spin over all "sect1" releases to find major release numbers
-sed -n 's/^ *<sect1 id="release-\([^-]-[^-]\).*/\1/p' "$FILE" |
-uniq |
-while read RELEASE
-do
-	# copy preamble
-	cp /tmp/preamble "$RELEASE"-"`basename $FILE`"
-	# grab remainder of file for major release
-	grep -A 10000000 "<sect1 id=\"release-$RELEASE" "$FILE" >> "$RELEASE"-"`basename $FILE`"
-done
-