8 files changed, 1000 insertions, 0 deletions
diff --git a/docs/internals/autowrite b/docs/internals/autowrite
new file mode 100644
index 0000000..dbad6c8
--- /dev/null
+++ b/docs/internals/autowrite
@@ -0,0 +1,88 @@
+#	@(#)autowrite	8.3 (Berkeley) 2/17/95
+
+Vi autowrite behavior, the fields with *'s are "don't cares".
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Commands that are affected only by autowrite:
+
+Command	File		Autowrite?	Action:
+	modified?
+-----------------------------------------------
+^Z	Y		Y		Write file and suspend.
+^Z	Y		N		Suspend.
+^Z	N		*		Suspend.
+
+# This behavior is NOT identical to :edit.
+^^	Y		Y		Write file and jump.
+^^	Y		N		Error.
+^^	N		*		Jump.
+
+# The new nvi command ^T (:tagpop) behaves identically to ^].
+# This behavior is identical to :tag, :tagpop, and :tagpush with
+# force always set to N.
+^]	Y		Y		Write file and jump.
+^]	Y		N		Error.
+^]	N		*		Jump.
+
+# There's no way to specify a force flag to the '!' command.
+:!	Y		Y		Write file and execute.
+:!	Y		N		Warn (if warn option) and execute.
+:!	N		*		Execute.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Commands that are affected by both autowrite and force:
+
+NOTE: the "force" flag is never passed on, i.e. the write
+to the file caused by the autowrite flag is never forced.
+
+Command	File		Autowrite?	Force? 	Action:
+	modified?			(!)
+-------------------------------------------------------
+# The first rule (YYY) is historic practice, but seems wrong.
+# In nvi, :next and :prev commands behave identically to :rewind.
+:next 	Y		Y		Y	Write changes and jump.
+:next 	Y		Y		N	Write changes and jump.
+:next 	Y		N		Y	Abandon changes and jump.
+:next 	Y		N		N	Error.
+:next 	N		*		*	Jump.
+
+:rewind	Y		Y		Y	Abandon changes and jump.
+:rewind	Y		Y		N	Write changes and jump.
+:rewind	Y		N		Y	Abandon changes and jump.
+:rewind	Y		N		N	Error.
+:rewind	N		*		*	Jump.
+
+# The new nvi commands, :tagpop and :tagtop, behave identically to :tag.
+# Note, this behavior is the same as :rewind and friends, as well.
+:tag	Y		Y		Y	Abandon changes and jump.
+:tag	Y		Y		N	Write changes and jump.
+:tag	Y		N		Y	Abandon changes and jump.
+:tag	Y		N		N	Error.
+:tag	N		*		*	Jump.
+
+# The command :suspend behaves identically to :stop.
+:stop	Y		Y		Y	Suspend.
+:stop	Y		Y		N	Write changes and suspend.
+:stop	Y		N		Y	Suspend.
+:stop	Y		N		N	Suspend.
+:stop	N		*		*	Suspend.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Commands that might be affected by autowrite, but aren't:
+
+Command	File		Autowrite?	Force? 	Action:
+	modified?			(!)
+-------------------------------------------------------
+#:ex, and :vi (executed while in vi mode) behave identically to :edit.
+:edit 	Y		*		Y	Abandon changes and jump.
+:edit 	Y		*		N	Error.
+:edit 	N		*		*	Jump.
+
+:quit	Y		*		Y	Quit.
+:quit	Y		*		N	Error.
+:quit	N		*		*	Quit.
+
+:shell	*		*		*	Execute shell.
+
+:xit	Y		*		*	Write changes and exit.
+:xit	N		*		*	Exit.
diff --git a/docs/internals/context b/docs/internals/context
new file mode 100644
index 0000000..8b1db32
--- /dev/null
+++ b/docs/internals/context
@@ -0,0 +1,32 @@
+#	@(#)context	8.6 (Berkeley) 10/14/94
+
+In historic vi, the previous context mark was always set:
+
+ex address:
+    any number, <question-mark>, <slash>, <dollar-sign>,
+    <single-quote>, <backslash>
+
+ex commands: undo, "z.", global, v
+
+vi commands: (, ), {, }, %, [[, ]], ^]
+
+nvi adds the vi command ^T to this list.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+In historic vi, the previous context mark was set if the
+line changed:
+
+vi commands: '<mark>, G, H, L, M, z
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+In historic vi, the previous context mark was set if the
+line or column changed:
+
+vi commands: `<mark>, /, ?, N, n
+
+nvi adds the vi command ^A to this list.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+In historic vi, the previous context mark was set in non-visual
+mode for ^R and ^L if the line changed, but I have yet to figure
+out how the line could change.
diff --git a/docs/internals/cscope.NOTES b/docs/internals/cscope.NOTES
new file mode 100644
index 0000000..e0e3483
--- /dev/null
+++ b/docs/internals/cscope.NOTES
@@ -0,0 +1,142 @@
+Cscope Notes:
+
+The nvi tags structure has been reworked to handle the notion of multiple
+locations per tag.  This supports cscope, which returns multiple locations
+per query.  It will hopefully support ctags programs that create databases
+with multiple locations per tag as well.
+
+There is now a list of "tag queues" chained from each screen.  Each tag
+queue has one or more "tag locations".
+
+   +----+    +----+	+----+     +----+
+   | EP | -> | Q1 | <-- | T1 | <-- | T2 |
+   +----+    +----+ --> +----+ --> +----+
+  	     |
+  	     +----+     +----+
+  	     | Q2 | <-- | T1 |
+  	     +----+ --> +----+
+  	     |
+  	     +----+	+----+
+  	     | Q3 | <-- | T1 |
+  	     +----+ --> +----+
+  
+In the above diagram, each "Q" is a "tag queue", and each "T" is a
+tag location.  Generally, the commands:
+
+	:tag			create a new Q
+	^[			create a new Q
+	:cscope find		create a new Q
+	:tagnext		move to the next T
+	:tagprev		move to the previous T
+	:tagpop			discard one or more Q's
+	^T			discard the most recent Q
+	:tagtop			discard all Q's
+
+More specifically:
+
+:cs[cope] a[dd] cscope-dir
+
+	Attach to the cscope database in cscope-dir.
+
+:cs[cope] f[ind] c|d|e|f|g|i|s|t buffer|pattern
+
+	Query all attached cscopes for the pattern.  The pattern is a
+	regular expression.  If the pattern is a double-quote character
+	followed by a valid buffer name (e.g., "t), then the contents
+	of the named buffer are used as the pattern.
+
+	c: find callers of name
+	d: find all function calls made from name
+	e: find pattern
+	f: find files with name as substring
+	g: find definition of name
+	i: find files #including name
+	s: find all uses of name
+	t: find assignments to name
+
+	The find command pushes the current location onto the tags stack,
+	and switches to the first location resulting from the query, if
+	the query returned at least one result.
+
+:cs[cope] h[elp] [command]
+	
+	List the cscope commands, or usage help on one command.
+
+:display c[onnections]
+
+	Display the list of cscope connections
+
+:display t[ags]
+
+	The tags display has been enhanced to display multiple tag
+	locations per tag query.
+
+:cs[cope] k[ill] #
+
+	Kill cscope connection number #.
+
+:cs[cope] r[eset]
+	Kill all attached cscopes.  Useful if one got hung but you don't
+	know which one.
+
+:tagn[ext][!]
+
+	Move to the next tag resulting from a query.
+
+:tagpr[ev][!]
+
+	Return to the previous tag resulting from a query.
+
+:tagp[op], ^T
+
+	Return to the previous tag group (no change).
+
+:tagt[op]
+
+	Discard all tag groups (no change).
+
+Suggested maps:
+
+	" ^N: move to the next tag
+	map ^N :tagnext^M
+	" ^P: move to the previous tag
+	map ^P :tagprev^M
+
+	" Tab+letter performs a C-Scope query on the current word.
+	" C-Scope 12.9 has a text-string query (type t).
+	" C-Scope 13.3 replaces it with an assignment query; hence a==t.
+	map <tab>a "tye:csc find t"t
+	map <tab>c "tye:csc find c"t
+	map <tab>d "tye:csc find d"t
+	map <tab>e "tye:csc find e"t
+	map <tab>f "tye:csc find f"t
+	map <tab>g "tye:csc find g"t
+	map <tab>i "tye:csc find i"t
+	map <tab>s "tye:csc find s"t
+	map <tab>t "tye:csc find t"t
+
+To start nvi with an initial set of cscope directories, use the environment
+variable CSCOPE_DIRS.  This variable should contain a <blank>-separated
+list of directories containing cscope databases.  (This MAY be changed to
+be an edit option, I haven't really decided, yet.)
+
+Each cscope directory must contain a file named "cscope.out" which is the
+main cscope database, or nvi will not attempt to connect to a cscope to
+handle requests for that database.
+
+The file "cscope.tpath" may contain a colon-separated directory search
+path which will be used to find the files reported by cscope.  If this
+cscope.tpath does not exist, then the paths are assumed to be relative to
+the cscope directory itself.  This is an extension to the standard cscope,
+but seems important enough to keep.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Cscope Availability:
+
+UNIXWare System V Release 4.0 variants such as Sun Solaris 2.x
+(/opt/SUNWspro/bin) have version 11.5, and UNIXWare System V
+Release 4.1 has version 12.10 with an option for much faster
+searching.
+
+You can buy version 13.3 source with an unrestricted license
+for $400 from AT&T Software Solutions by calling +1-800-462-8146.
diff --git a/docs/internals/gdb.script b/docs/internals/gdb.script
new file mode 100644
index 0000000..a112234
--- /dev/null
+++ b/docs/internals/gdb.script
@@ -0,0 +1,76 @@
+#	@(#)gdb.script	8.5 (Berkeley) 5/4/96
+
+# display the VI screen map
+# usage dmap(sp)
+define	dmap
+	set $h = ((VI_PRIVATE *)$arg0->vi_private)->h_smap
+	set $t = ((VI_PRIVATE *)$arg0->vi_private)->t_smap
+	while ($h <= $t)
+		printf "lno: %2d; soff %d coff %d ", \
+		    (int)$h->lno, (int)$h->soff, (int)$h->coff
+		if ($h->c_ecsize == 0)
+			printf "flushed\n"
+		else
+			printf "\n\tsboff %d; scoff %d\n", \
+			    (int)$h->c_sboff, (int)$h->c_scoff
+			printf "\teboff %d; eclen %d; ecsize %d\n", \
+			    (int)$h->c_eboff, (int)$h->c_eclen, \
+			    (int)$h->c_ecsize
+		end
+		set $h = $h + 1
+	end
+end
+
+# display the tail of the VI screen map
+define	tmap
+	set $h = ((VI_PRIVATE *)$arg0->vi_private)->h_smap
+	set $t = ((VI_PRIVATE *)$arg0->vi_private)->t_smap
+	while ($t >= $h)
+		printf "lno: %2d; soff %d coff %d ", \
+		    (int)$t->lno, (int)$t->soff, (int)$t->coff
+		if ($t->c_ecsize == 0)
+			printf "flushed\n"
+		else
+			printf "\n\tsboff %d; scoff %d\n", \
+			    (int)$t->c_sboff, (int)$t->c_scoff
+			printf "\teboff %d; eclen %d; ecsize %d\n", \
+			    (int)$t->c_eboff, (int)$t->c_eclen, \
+			    (int)$t->c_ecsize
+		end
+		set $t = $t - 1
+	end
+end
+
+# display the private structures
+define	clp
+	print *((CL_PRIVATE *)sp->gp->cl_private)
+end
+define	vip
+	print *((VI_PRIVATE *)sp->vi_private)
+end
+define	exp
+	print *((EX_PRIVATE *)sp->ex_private)
+end
+
+# display the marks
+define	markp
+	set $h = sp->ep->marks.next
+	set $t = &sp->ep->marks
+	while ($h != 0 && $h != $t)
+		printf "key %c lno: %d cno: %d flags: %x\n", \
+		    ((MARK *)$h)->name, ((MARK *)$h)->lno, \
+		    ((MARK *)$h)->cno, ((MARK *)$h)->flags
+		set $h = ((MARK *)$h)->next
+	end
+end
+
+# display the tags
+define	tagp
+	set $h = sp->taghdr.next
+	set $t = &sp->taghdr
+	while ($h != 0 && $h != $t)
+		printf "tag: %s lno %d cno %d\n", ((TAG *)$h)->frp->fname, \
+		    ((TAG *)$h)->lno, ((TAG *)$h)->cno
+		set $h= ((TAG *)$h)->next
+	end
+end
diff --git a/docs/internals/input b/docs/internals/input
new file mode 100644
index 0000000..9a7506e
--- /dev/null
+++ b/docs/internals/input
@@ -0,0 +1,350 @@
+#	@(#)input	5.5 (Berkeley) 7/2/94
+
+MAPS, EXECUTABLE BUFFERS AND INPUT IN EX/VI:
+
+The basic rule is that input in ex/vi is a stack.  Every time a key which
+gets expanded is encountered, it is expanded and the expansion is treated
+as if it were input from the user.  So, maps and executable buffers are
+simply pushed onto the stack from which keys are returned.  The exception
+is that if the "remap" option is turned off, only a single map expansion
+is done.  I intend to be fully backward compatible with this.
+
+Historically, if the mode of the editor changed (ex to vi or vice versa),
+any queued input was silently discarded.  I don't see any reason to either
+support or not support this semantic.  I intend to retain the queued input,
+mostly because it's simpler than throwing it away.
+
+Historically, neither the initial command on the command line (the + flag)
+or the +cmd associated with the ex and edit commands was subject to mapping.
+Also, while the +cmd appears to be subject to "@buffer" expansion, once
+expanded it doesn't appear to work correctly.  I don't see any reason to
+either support or not support these semantics, so, for consistency, I intend
+to pass both the initial command and the command associated with ex and edit
+commands through the standard mapping and @ buffer expansion.
+
+One other difference between the historic ex/vi and nex/nvi is that nex
+displays the executed buffers as it executes them.  This means that if
+the file is:
+
+	set term=xterm
+	set term=yterm
+	set term=yterm
+
+the user will see the following during a typical edit session:
+
+	nex testfile
+	testfile: unmodified: line 3
+	:1,$yank a
+	:@a
+	:set term=zterm
+	:set term=yterm
+	:set term=xterm
+	:q!
+
+This seems like a feature and unlikely to break anything, so I don't
+intend to match historic practice in this area.
+
+The rest of this document is a set of conclusions as to how I believe
+the historic maps and @ buffers work.  The summary is as follows:
+
+1: For buffers that are cut in "line mode", or buffers that are not cut
+   in line mode but which contain portions of more than a single line, a
+   trailing <newline> character appears in the input for each line in the
+   buffer when it is executed.  For buffers not cut in line mode and which
+   contain portions of only a single line, no additional characters
+   appear in the input.
+2: Executable buffers that execute other buffers don't load their
+   contents until they execute them.
+3: Maps and executable buffers are copied when they are executed --
+   they can be modified by the command but that does not change their
+   actions.
+4: Historically, executable buffers are discarded if the editor
+   switches between ex and vi modes.
+5: Executable buffers inside of map commands are expanded normally.
+   Maps inside of executable buffers are expanded normally.
+6: If an error is encountered while executing a mapped command or buffer,
+   the rest of the mapped command/buffer is discarded.  No user input
+   characters are discarded.
+7: Characters in executable buffers are remapped.
+8: Characters in executable buffers are not quoted.
+
+Individual test cases follow.  Note, in the test cases, control characters
+are not literal and will have to be replaced to make the test cases work.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+1: For buffers that are cut in "line mode", or buffers that are not cut
+   in line mode but which contain portions of more than a single line, a
+   trailing <newline> character appears in the input for each line in the
+   buffer when it is executed.  For buffers not cut in line mode and which
+   contain portions of only a single line, no additional characters
+   appear in the input.
+
+===   test file   ===
+3Gw
+w
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+=== end test file ===
+
+   If the first line is loaded into 'a' and executed:
+
+1G"ayy@a
+
+   The cursor ends up on the '2', a result of pushing "3Gw^J" onto
+   the stack.
+
+   If the first two lines are loaded into 'a' and executed:
+
+1G2"ayy@a
+
+   The cursor ends up on the 'f' in "foo" in the fifth line of the
+   file, a result of pushing "3Gw^Jw^J" onto the stack.
+
+   If the first line is loaded into 'a', but not using line mode,
+   and executed:
+
+1G"ay$@a
+
+   The cursor ends up on the '1', a result of pushing "3Gw" onto
+   the stack
+
+   If the first two lines are loaded into 'a', but not using line mode,
+   and executed:
+
+1G2"ay$@a
+
+   The cursor ends up on the 'f' in "foo" in the fifth line of the
+   file, a result of pushing "3Gw^Jw^J" onto the stack.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+2: Executable buffers that execute other buffers don't load their
+   contents until they execute them.
+
+===   test file   ===
+cwLOAD B^[
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+@a@b
+"byy
+=== end test file ===
+
+   The command is loaded into 'e', and then executed.  'e' executes
+   'a', which loads 'b', then 'e' executes 'b'.
+
+5G"eyy6G"ayy1G@e
+
+   The output should be:
+
+===   output file   ===
+cwLOAD B^[
+LOAD B 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+@a@b
+"byy
+=== end output file ===
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+3: Maps and executable buffers are copied when they are executed --
+   they can be modified by the command but that does not change their
+   actions.
+
+   Executable buffers:
+
+===   test file   ===
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+@a@b
+"eyy
+cwEXECUTE B^[
+=== end test file ===
+
+4G"eyy5G"ayy6G"byy1G@eG"ep
+
+   The command is loaded into 'e', and then executed.  'e' executes
+   'a', which loads 'e', then 'e' executes 'b' anyway.
+
+   The output should be:
+
+===   output file   ===
+line 1 foo bar baz
+EXECUTE B 2 foo bar baz
+line 3 foo bar baz
+@a@b
+"eyy
+cwEXECUTE B^[
+line 1 foo bar baz
+=== end output file ===
+
+   Maps:
+
+===   test file   ===
+Cine 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+=== end test file ===
+
+   Entering the command ':map = :map = rB^V^MrA^M1G==' shows that
+   the first time the '=' is entered the '=' map is set and the
+   character is changed to 'A', the second time the character is
+   changed to 'B'.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+4: Historically, executable buffers are discarded if the editor
+   switches between ex and vi modes.
+
+===   test file   ===
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+cwCHANGE^[Q:set
+set|visual|1Gwww
+=== end test file ===
+
+vi testfile
+4G"ayy@a
+
+ex testfile
+$p
+yank a
+@a
+
+   In vi, the command is loaded into 'a' and then executed.  The command
+   subsequent to the 'Q' is (historically, silently) discarded.
+
+   In ex, the command is loaded into 'a' and then executed.  The command
+   subsequent to the 'visual' is (historically, silently) discarded.  The
+   first set command is output by ex, although refreshing the screen usually
+   causes it not to be seen.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+5: Executable buffers inside of map commands are expanded normally.
+   Maps inside of executable buffers are expanded normally.
+
+   Buffers inside of map commands:
+
+===   test file   ===
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+cwREPLACE BY A^[
+=== end test file ===
+
+4G"ay$:map x @a
+1Gx
+
+   The output should be:
+
+===   output file   ===
+REPLACE BY A 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+cwREPLACE BY A^[
+=== end output file ===
+
+   Maps commands inside of executable buffers:
+
+===   test file   ===
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+X
+=== end test file ===
+
+:map X cwREPLACE BY XMAP^[
+4G"ay$1G@a
+
+   The output should be:
+
+===   output file   ===
+REPLACE BY XMAP 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+X
+=== end output file ===
+
+   Here's a test that does both, repeatedly.
+
+===   test file   ===
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+X
+Y
+cwREPLACED BY C^[
+blank line
+=== end test file ===
+
+:map x @a
+4G"ay$
+:map X @b
+5G"by$
+:map Y @c
+6G"cy$
+1Gx
+
+   The output should be:
+
+===   output file   ===
+REPLACED BY C 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+X
+Y
+cwREPLACED BY C^[
+blank line
+=== end output file ===
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+6: If an error is encountered while executing a mapped command or
+   a buffer, the rest of the mapped command/buffer is discarded.  No
+   user input characters are discarded.
+
+===   test file   ===
+line 1 foo bar baz
+line 2 foo bar baz
+line 3 foo bar baz
+:map = 10GcwREPLACMENT^V^[^[
+=== end test file ===
+
+   The above mapping fails, however, if the 10G is changed to 1, 2,
+   or 3G, it will succeed.
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+7: Characters in executable buffers are remapped.
+
+===   test file   ===
+abcdefghijklmnnop
+ggg
+=== end test file ===
+
+:map g x
+2G"ay$1G@a
+
+   The output should be:
+
+===   output file   ===
+defghijklmnnop
+ggg
+=== end output file ===
+
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+8: Characters in executable buffers are not quoted.
+
+===   test file   ===
+iFOO^[
+
+=== end test file ===
+
+1G"ay$2G@a
+
+   The output should be:
+
+===   output file   ===
+iFOO^[
+FOO
+=== end output file ===
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
diff --git a/docs/internals/openmode b/docs/internals/openmode
new file mode 100644
index 0000000..c64b767
--- /dev/null
+++ b/docs/internals/openmode
@@ -0,0 +1,36 @@
+	@(#)openmode	8.1 (Berkeley) 10/29/94
+
+Open mode has the following special behaviors:
+
+z, ^F, ^B:
+	If count is not specified, it shall default to the window
+	edit option - 2.
+
+	Write lines from the edit buffer starting at:
+
+		(the current line) - ((count - 2) / 2)
+
+	until:
+
+		(((count + 1) / 2) * 2) - 1
+	
+	lines, or the last line in the edit buffer has been written.  A
+	line consisting of the smaller of the number of columns in the
+	display divided by two or 40 ``-'' characters shall be written
+	immediately before and after the specified is written.  These two
+	lines shall count against the total number of lines to be written.
+	A blank line shall be written after the last line is written.
+
+	z, ^F and ^B all behave identically.
+
+^D:	Display the next scroll value lines, change the current line.
+
+^U:	Change the current line, do nothing else.
+
+^E, ^Y:	Do nothing.
+
+^L:	Clear the screen and redisplay the current line.
+
+H, L, M:
+	Move to the first nonblank of the current line and do nothing
+	else.
diff --git a/docs/internals/quoting b/docs/internals/quoting
new file mode 100644
index 0000000..a5fb892
--- /dev/null
+++ b/docs/internals/quoting
@@ -0,0 +1,208 @@
+#	@(#)quoting	5.5 (Berkeley) 11/12/94
+
+QUOTING IN EX/VI:
+
+There are four escape characters in historic ex/vi:
+
+	\ (backslashes)
+	^V
+	^Q (assuming it wasn't used for IXON/IXOFF)
+	The terminal literal next character.
+
+Vi did not use the lnext character, it always used ^V (or ^Q).
+^V and ^Q were equivalent in all cases for vi.
+
+There are four different areas in ex/vi where escaping characters
+is interesting:
+
+	1: In vi text input mode.
+	2: In vi command mode.
+	3: In ex command and text input modes.
+	4: In the ex commands themselves.
+
+1: Vi text input mode (a, i, o, :colon commands, etc.):
+
+   The set of characters that users might want to escape are as follows.
+   As ^L and ^Z were not special in input mode, they are not listed.
+
+	carriage return (^M)
+	escape		(^[)
+	autoindents	(^D, 0, ^, ^T)
+	erase		(^H)
+	word erase	(^W)
+	line erase	(^U)
+	newline		(^J)		(not historic practice)
+
+   Historic practice was that ^V was the only way to escape any
+   of these characters, and that whatever character followed
+   the ^V was taken literally, e.g. ^V^V is a single ^V.  I
+   don't see any strong reason to make it possible to escape
+   ^J, so I'm going to leave that alone.
+
+   One comment regarding the autoindent characters.  In historic
+   vi, if you entered "^V0^D" autoindent erasure was still
+   triggered, although it wasn't if you entered "0^V^D".  In
+   nvi, if you escape either character, autoindent erasure is
+   not triggered.
+
+   Abbreviations were not performed if the non-word character
+   that triggered the abbreviation was escaped by a ^V.  Input
+   maps were not triggered if any part of the map was escaped
+   by a ^V.
+
+   The historic vi implementation for the 'r' command requires
+   two leading ^V's to replace a character with a literal
+   character.  This is obviously a bug, and should be fixed.
+
+2: Vi command mode
+
+   Command maps were not triggered if the second or later
+   character of a map was escaped by a ^V.
+
+   The obvious extension is that ^V should keep the next command
+   character from being mapped, so you can do ":map x xxx" and
+   then enter ^Vx to delete a single character.
+
+3: Ex command and text input modes.
+
+   As ex ran in canonical mode, there was little work that it
+   needed to do for quoting.  The notable differences between
+   ex and vi are that it was possible to escape a <newline> in
+   the ex command and text input modes, and ex used the "literal
+   next" character, not control-V/control-Q.
+
+4: The ex commands:
+
+   Ex commands are delimited by '|' or newline characters.
+   Within the commands, whitespace characters delimit the
+   arguments.  Backslash will generally escape any following
+   character.  In the abbreviate, unabbreviate, map and unmap
+   commands, control-V escapes the next character, instead.
+
+   This is historic behavior in vi, although there are special
+   cases where it's impossible to escape a character, generally
+   a whitespace character.
+
+   Escaping characters in file names in ex commands:
+
+	:cd [directory]				(directory)
+	:chdir [directory]			(directory)
+	:edit [+cmd] [file]			(file)
+	:ex [+cmd] [file]			(file)
+	:file [file]				(file)
+	:next [file ...]			(file ...)
+	:read [!cmd | file]			(file)
+	:source [file]				(file)
+	:write [!cmd | file]			(file)
+	:wq [file]				(file)
+	:xit [file]				(file)
+
+   Since file names are also subject to word expansion, the
+   underlying shell had better be doing the correct backslash
+   escaping.  This is NOT historic behavior in vi, making it
+   impossible to insert a whitespace, newline or carriage return
+   character into a file name.
+
+4: Escaping characters in non-file arguments in ex commands:
+
+	:abbreviate word string			(word, string)
+*	:edit [+cmd] [file]			(+cmd)
+*	:ex [+cmd] [file]			(+cmd)
+	:map word string			(word, string)
+*	:set [option ...]			(option)
+*	:tag string				(string)
+	:unabbreviate word			(word)
+	:unmap word				(word)
+
+   These commands use whitespace to delimit their arguments, and use
+   ^V to escape those characters.  The exceptions are starred in the
+   above list, and are discussed below.
+
+   In general, I intend to treat a ^V in any argument, followed by
+   any character, as that literal character.  This will permit
+   editing of files name "foo|", for example, by using the string
+   "foo\^V|", where the literal next character protects the pipe
+   from the ex command parser and the backslash protects it from the
+   shell expansion.
+
+   This is backward compatible with historical vi, although there
+   were a number of special cases where vi wasn't consistent.
+
+4.1: The edit/ex commands:
+
+   The edit/ex commands are a special case because | symbols may
+   occur in the "+cmd" field, for example:
+
+	:edit +10|s/abc/ABC/ file.c
+
+   In addition, the edit and ex commands have historically
+   ignored literal next characters in the +cmd string, so that
+   the following command won't work.
+
+	:edit +10|s/X/^V / file.c
+
+   I intend to handle the literal next character in edit/ex consistently
+   with how it is handled in other commands.
+
+   More fun facts to know and tell:
+	The acid test for the ex/edit commands:
+
+		date > file1; date > file2
+		vi
+		:edit +1|s/./XXX/|w file1| e file2|1 | s/./XXX/|wq
+
+	No version of vi, of which I'm aware, handles it.
+
+4.2: The set command:
+
+   The set command treats ^V's as literal characters, so the
+   following command won't work.  Backslashes do work in this
+   case, though, so the second version of the command does work.
+
+	set tags=tags_file1^V tags_file2
+	set tags=tags_file1\ tags_file2
+
+   I intend to continue permitting backslashes in set commands,
+   but to also permit literal next characters to work as well.
+   This is backward compatible, but will also make set
+   consistent with the other commands.  I think it's unlikely
+   to break any historic .exrc's, given that there are probably
+   very few files with ^V's in their name.
+
+4.3: The tag command:
+
+   The tag command ignores ^V's and backslashes; there's no way to
+   get a space into a tag name.
+
+   I think this is a don't care, and I don't intend to fix it.
+
+5: Regular expressions:
+
+	:global /pattern/ command
+	:substitute /pattern/replace/
+	:vglobal /pattern/ command
+
+   I intend to treat a backslash in the pattern, followed by the
+   delimiter character or a backslash, as that literal character.
+
+   This is historic behavior in vi.  It would get rid of a fairly
+   hard-to-explain special case if we could just use the character
+   immediately following the backslash in all cases, or, if we
+   changed nvi to permit using the literal next character as a
+   pattern escape character, but that would probably break historic
+   scripts.
+
+   There is an additional escaping issue for regular expressions.
+   Within the pattern and replacement, the '|' character did not
+   delimit ex commands.  For example, the following is legal.
+
+	:substitute /|/PIPE/|s/P/XXX/
+
+   This is a special case that I will support.
+
+6: Ending anything with an escape character:
+
+   In all of the above rules, an escape character (either ^V or a
+   backslash) at the end of an argument or file name is not handled
+   specially, but used as a literal character.
+
diff --git a/docs/internals/structures b/docs/internals/structures
new file mode 100644
index 0000000..a25c780
--- /dev/null
+++ b/docs/internals/structures
@@ -0,0 +1,68 @@
+#	@(#)structures	5.4 (Berkeley) 10/4/95
+
+There are three major data structures in this package, plus a single data
+structure per screen type.  The first is a single global structure (GS)
+which contains information common to all files and screens.  It hold
+global things like the input key queues, and functions as a single place
+to hang things.  For example, interrupt routines have to be able to find
+screen structures, and they can only do this if they have a starting
+point.  The number of globals in nvi is dependent on the screen type, but
+every screen type will have at least one global, __global_list, which
+references the GS structure.
+
+The GS structure contains linked lists of screen (SCR) structures.
+Each SCR structure normally references a file (EXF) structure.
+
+The GS structure has a set of functions which update the screen and/or
+return information about the screen from the underlying screen package.
+The GS structure never goes away.  The SCR structure persists over
+instances of screens, and the EXF structure persists over references to
+files.
+
+File names have different properties than files themselves, so the name
+information for a file is held in an FREF structure which is chained from
+the SCR structure.
+
+In general, functions are always passed an SCR structure, which usually
+references an underlying EXF structure.  The SCR structure is necessary
+for any routine that wishes to talk to the screen, the EXF structure is
+necessary for any routine that wants to modify the file.  The relationship
+between an SCR structure and its underlying EXF structure is not fixed,
+and various ex commands will substitute a new EXF in place of the current
+one, and there's no way to detect this.
+
+The naming of the structures is consistent across the program.  (Macros
+even depend on it, so don't try and change it!)  The global structure is
+"gp", the screen structure is "sp", and the file structure is "ep".
+
+A few other data structures:
+
+TEXT	In nvi/cut.h.  This structure describes a portion of a line,
+	and is used by the input routines and as the "line" part of a
+	cut buffer.
+
+CB	In nvi/cut.h.	A cut buffer.  A cut buffer is a place to
+	hang a list of TEXT structures.
+
+CL	The curses screen private data structure.  Everything to
+	do standalone curses screens.
+
+MARK	In nvi/mark.h.  A cursor position, consisting of a line number
+	and a column number.
+
+MSG	In nvi/msg.h.  A chain of messages for the user.
+
+SEQ	In nvi/seq.h.  An abbreviation or a map entry.
+
+TK	The Tcl/Tk screen private data structure.  Everything to
+	do standalone Tcl/Tk screens.
+
+EXCMD   In nvi/ex/ex.h.  The structure that gets passed around to the
+	functions that implement the ex commands.  (The main ex command
+	loop (see nvi/ex/ex.c) builds this up and then passes it to the
+	ex functions.)
+
+VICMD	In nvi/vi/vi.h.  The structure that gets passed around to the
+	functions that implement the vi commands.  (The main vi command
+	loop (see nvi/vi/vi.c) builds this up and then passes it to the
+	vi functions.)