qasm/IIGS.601.SRC/Debugger/Old.Debugger.Comments.1988

;
;
; May 11, 1988   	1)  Converted the debugger to run
;	in ROM.  On power-up the debugger
;	looks like a DA and then the system
;	copies itself into RAM (entire 64K bank)
;	It then steals the DA vector and then
;	replaces the starting address of the debugger
;	in its place.  Once the debugger has been invoked,
;	it asks for 1K of bank zero memory for stack and
;	direct page use.  It saves the environment of any
;	memory that it uses.  It presently uses the DA 
;	calls to save and restore the text screens for exit
;
;	2)  Cleaned up several bugs that had been present 
;	from the beginning of time:
;
;	A)  The debugger screwed up the stack when it
;	transferred from native mode to emulation mode
;	and back.  I saved the upper half of the native
;	mode stack pointer and restore it when going
;	to native.
;
;	B)  When loading files to debug, the system loader
;	sets aside a portion of zero page to hold some
;	variables it needs.  When the debugger quit
;	it did not dispose of this memory.  This made
;	it impossible for ProDOS 8 applications to
;	run following the debugger because they reserve
;	all of bank zero.  I saved all of the file ID's
;	and then dispose of them and all of their memory
;	when I quit.  This also enabled people to
;	load more than one file at a time and still
;	keep track of all the memory.  15 max files
;	can be loaded.
;
;	3)  Added the feature to allow JSR's to be run real
;	time.  Before only JSL's could.  Now this only
;	works in RAM.
;
;	4)  Fixed the re-entrancy problem.  Once you are inside
;	the debugger, attempts to access it again are useless.
;	The only way to re-enter is to quit and then get
;	back in using CTRL-OA-OPTION-ESC.
;
; May 22,1988	1)  Stored the current stack in 01/0100 on exit.  To keep
;	interrupts from trashing my input/output variables,
;	I pushed them on entrant stack and then saved stack
;	in 01/0100.  Problem was that stack was drifting down
;	every time the debugger was called about 1B bytes and
;	quickly drifted into zero page and died!!.  I'm now
;	saving the stack on exit.  Now the stack drifts three
;	bytes every time (from rtl that follows).  Will check and
;	see if that should be fixed.
;
;	2)  saved and restored the graphics switches -- including
;	the super hi-res switch.  Now you can return to the finder
;	and see the pretty screen.  Only works if you have not
;	used the graphics screens -- I'm still thinking about 
;	that!!
;
; May 23,1988	1)  When a break occurred, the break handler was still set
;	to the ROM bank.  This was changed to be in the RAM bank.
;
;	2)  When the jump (J) command was used, the proper bank was
;	not saved in MYID to be used to get the proper address
;	to jump to like it is in the JSL and JSR.  This was fixed.
;
;	3)  A problem was occurring with the heart beat task.  If the
;	DA vector had not changed from the last task and was
;	still my vector, then the task replaced the CDA vector
;	with my vector.  CDA's were never available and trying
;	to get at one hung the system in a wait loop.
;
;	4)  Added three more to the stack before restoring it in QUIT.
;	This insures that the stack pointer is in the EXACT same
;	place as when it entered the debugger.  Should make 
;	everyone happy.!!
;
;	5)  The stack is still causing me problems.  This time I'm
;	saving the emulation stack pointer away and then restoring
;	it when I leave.  This way it is always exactly the same.
;
; May 26,1988	1)  Fixed subtle stack problem.  The lower byte of the A reg
;	wasn't being saved in the save loop due to a bne that
;	should have been a bpl.  If the debugger was called
;	in between an instruction that loaded A and then 
;	transferred it to the stack, the stack got trashed because
;	the proper value wasn't being restored.
;
; May 27,1988	1)  Added the code that allows real time JSR's to be run
;	out of ROM!! The code keeps a table for each ROM bank
;	The first time a ROM JSR is called, the code searches the
;	bank for any RTL and uses that address for the RTS to
;	return to.  The next time a JSR is needed in the same
;	bank, it has it saved and doesn't need to look for it.
;	Now there is no provision for what happens if there is
;	no RTL anywhere in the bank.
;
; May 31,1988	1)  Went through and added all of the longa and longi's to
;	go with the rep's and sep's in the code.  Did it care-
;	fully to avoid any crashes when the code is assembled
;	and run.
;
; June 1,1988	1)  Made a special ROM for the people who changed the CDA 
;	vector.  The normal one is E1/0048, their's is E1/02B4
;
; June 2,1988	1)  One good hack deserves another.  For Fern's special ROM's
;	I set up the CDA vector to point to the manager -- like it
;	supposedly is on power-up.  This insures that you can 
;	always get into the DA's
;
;	2)  Changed the save and restore screen calls from the DA one
;	to the one that requests a handle and then puts all of the
;	memory there.  There is no provision for error handling yet.  
;	This allows DA's to be called from the debugger.
;
; June 8,1988	1)  There is a problem with the boot code turning off the 
;	VBL's. This caused problems because the heartbeat task 
;	needed to keep me installed was not being called.  Thus, 
;	the hack:  when the debugger is called, I re-install my 
;	heartbeat task.  If it is already there, I simply turn on 
;	VBL's and go for it.  If it is not there, then I assume 
;	that a pr#5 or something similar wiped me out of the 
;	heartbeat queue.  I install myself, clear out my queue and 
;	turn on VBL's.  With this solution, the worst thing that 
;	can happen, is that someone will try to get into DA's, find
;	they can't, access the DA's.
;
; June 8,1988	1)  Changed the base conversion feature to handle expressions of 
;	32-bit numbers. An expression consists of a series of one or
;	more hex or decimal numbers separated by operators and 
;	ending with an equal sign. Hex numbers must be preceded by 
;	a '$'.  Decimal numbers may be preceded by a positive or
;	negative sign or nothing.  Negative numbers are converted 
;	to 32-bit two's-complement.  Numbers that are larger that 
;	32-bits will wrap. Legal operators are '+', '-','*', '/', 
;	and '%' (modulus).  All operators are evaluated left to 
;	right with equal precedence. Overflow is ignored. The result 
;	will be printed in either hex or decimal depending on the 
;	form of the expression. If the expression consists of just one 
;	hex number, then it is assumed we want to convert it to 
;	decimal, so the result will be in decimal.  If the expression
;	is just one decimal number, the result is printed in hex. 
;	If the expression contains more than one number and all the 
;	numbers are in decimal,then the result is printed in decimal.  
;	Otherwise the result is printed in hex.
;
;	2)  Changed command parser to read command starting at the 
;	rightmost equal sign in the current line that is not at the 
;	end of the line.  This allows the result of a calculation to 
;	be used in a command without having to retype the result.
;
; June 9, 1988	1)  Added indirect memory referencing.  An address followed by two 
;	colons will dump memory starting at the 3-byte address which 
;	is stored at the address given in the command.
;	
;	2)  Added direct page mode for the memory subdisplay.  Typing 'Z' 
;	while editing a memory cell will toggle direct page mode.  
;	A direct page memory cell will display the given direct page 
;	address even when the direct page changes.
;
; June 10, 1988	1)  Added a new disassembler mode that causes the disassembler to
;	recognize REP's and SEP's and switches between long and short 
;	accumulator and index registers automatically during 
;	disassembly.The mode is controlled by the disassembly mode 
;	flag (d).  If d is set, then SEP's and REP's are recognized; 
;	if d is clear, then they are ignored.  The flag can be set or 
;	cleared by the command d=x, where x is either 0 or 1.
;
; June 13, 1988	1)  Added dereferencing mode for the memory subdisplay.  Typing 
;	'I' while editing a memory cell will toggle deref mode.
;	Two colons are used to indicate a deref.  The memory 
;	displayed will be the contents of the memory location whose
;	address is stored in the three bytes starting at the address
;	of the memory cell.
;
; June 14, 1988	1)  Added alignment of bytes in the dump display, so that the 
;	position of the byte is determined by the low nibble of it's 
;	address.
;
;	2)  Added a dump range option.  The command 'addr1.addr2:' dumps
;	memory between and including addr1 and addr2.  If the number
;	of bytes requested is greater than the number that will fit 
;	on the screen, then only what will fit on the screen is 
;	displayed.
;
; June 15, 1988	1)  The deref memory dump option was extended so that the number
;	of colons following the dump address or range is the number of
;	bytes in the indirect address.  That is the command of the 
;	form '$XXXXXX::' will dump the block of memory starting at 
;	the two byte address stored at $XXXXXX.  The bank is decided
;	by the value of the bank register.  A command of the form 
;	'$XXXXXX:::' will dump the block of memory starting at the 
;	three byte address stored at $XXXXXX.
;
;	2)  The memory subdisplay was updated to handle both 16-bit and
;	24-bit dereferencing.  The referencing is done as described 
;	for the dumping. The type of referencing (16-bit, 24-bit, or 
;	none) is changed by typing a colon while editing a memory 
;	cell.  The current referencing mode is indicated by the 
;	number of colons following the address as described above.
;
;	3)  Allowed entering of addresses into memory.  When setting
;	memory, if the value is longer than 1 byte, it is assumed 
;	to be an address. For example, typing '00/2000:123456' will
;	store a $56 at 00/2000, a $34 at 00/2001, and a $12 at 
;	00/2002.  Note that if the address is a 24-bit address, 
;	you must explicitly enter leading zero's.If you want to 
;	store the 24-bit address 00/1234 at 00/2000, you must type 
;	'00/2000:001234'.  If you type '00/2000:1234', it will be 
;	assumed to be a 16-bit address, and only two bytes will be 
;	entered into memory.  Note that the v1.0d1 ERS (5/11/88) 
;	implies that this feature already exists.  In fact, 
;	it did not exist.
;
;	4)  Added a quickdump feature.  If you type an address and then
;	press return without following it with a colon, then the 16 
;	bytes starting at that address are displayed at the bottom 
;	of the screen.  They will stay there until any key is pressed.
;	This keypress will be passed to the input routine, so it will 
;	be the first character of the next command.
;
; June 16, 1988	1)  Changed the expression routines so that decimal results are 
;	preceded by a plus.  This makes sure that they are not 
;	confused with hex results.
;
;	2)  Added a column heading line to the dump display.
;
; June 20, 1988	1)  A circular history queue was added to record the last 32 
;	instructions in a trace.  This queue is used to show the 
;	trace history when returning from tracing with the display
;	off. This means that a fast trace can be done by turning 
;	the display off first, but that you can still see the 
;	instructions that have been executed when done tracing.
;
; June 21, 1988	1)  Added display mode switches to the command mode.  These 
;	switches are similar to the ones available in trace mode.  
;	The backslash key '\' is now treated specially.  If this 
;	key is pressed any time while on the command line, then 
;	the next key pressed will be interpreted as a display switch
;	key.  The display switch keys are the same as in trace
;	mode.  Neither the backslash or the switch key will be 
;	displayed, so they can be entered in the middle of a command
;	without affecting it.
;
; June 22, 1988	1)  Initial register values are set to the saved irq values 
;	found in the $E1000 bank.
;
; June 23, 1988	1)  Removed previous change (hopefully temporarily).
;
; June 27, 1988	1)  Fixed trace history to work with stepping and with tracing
;	using the 'T' command without an address.
;
; June 28, 1988	1)  Fixed 80COLSTORE switch bug.  The debugger now restores the
;	value of the switch after it clears it.
;
; June 30, 1988	1)  Added conditional assembly stuff to allow the creation of
;	an init version of the debugger.  This version should be
;	made into a file type of $B6 on the GS.  It will load on
;	bootup and act exactly like the rom version.
;
;	2)  Updated the copyright/version message.
;
; July 5, 1988	1)  Removed code that initialized the display screen (i.e.
;	break points, mem protection, mem cells, stack display,
;	etc.)  The initial values are now determined by the
;	values in the various tables and variables at assembly
;	time.  Doing this means that if the values are changed
;	when in the desk accessory version, then if we exit the
;	debugger and then return, the values will be the same.
;
;	2)  Fixed bug in quick dump routine that caused crash when
;	exiting the debugger.
;
;	3)  Modified code so that the entry state is now saved as
;	variables instead of on the stack.
;
;	4)  No longer return to calling procedure with a rtl.  Now
;	we pull the return address off of the stack on entry
;	and do a jmp to that address to exit.
;
;	5)  The return address and processor status that were saved
;	on the user's stack by the interrupt routines are now
;	saved on entry to the debugger, and then restored on
;	exit.  This allows the user to trace his/her program
;	in the debugger without having to worry about destroying
;	the return stuff on the stack.
;
; July 6, 1988	1)  On initial startup, the break vector is patched to jump
;	into the debugger.  When a break in encountered, the
;	the debugger will launch and check the break status flag.
;	If it is set to debugger breaks, then the debugger will
;	beep and then set all of the registers to reflect the
;	status of the machine at the time of the break.  The stack
;	pointer will be set to it's value at the time of the break
;	(not after the break,) and the PC will be set to the
;	address of the break instruction.  If the break status
;	flag is set to use breaks, then the the debugger will just
;	break to the monitor.
;
;	2)  When entering the debugger by using the entry key
;	sequence, debugger will reflect the state of the machine
;	when the interrupt occurred.
;
;	3)  Fixed bug that caused crash when doing a J or an X with
;	either m=1 or e=1.  The fix involved restoring some lines
;	that had been commented out in the jump routine (JPC.)
;
;	4)  Added new debugger command 'START' that prints out the
;	address at which the debugger is loaded.  This is to
;	help in the debugging of the debugger.  This command is
;	only available if the debugger is assembled with
;	ThisIsDEV equal to 1.
;
; July 7, 1988	1)  Fixed bug in the routine that check to see if the break
;	vector has already been patched.  It now compares the
;	bank in the break vector against the current bank, not
;	the bank at assembly (or load) time.  This was only a
;	problem in the rom version.
;
; July 8, 1988	1)  Fixed bug that that caused problems when executing an JSL
;	or JSR at real time with the emulation mode on.
;
; 	2)  Fixed bug that caused the high byte of both X and Y to be
;	trashed when returning from executing a JSL or JSR at real
;	time.
;
;	3)  Fixed bug in 'X' command that sometimes caused a jump to
;	the wrong bank.
;
;	4)  Fixed bug that caused a crash when tracing a JSR in real
;	time with short accumulator.
;
;	5)  Added two new commands: STACK and DP.  These commands
;	respectively set the users stack pointer and direct page
;	pointer to point to the stack and direct page that are
;	allocated by the debugger on startup.  These are primarily
;	useful for when the user enters the non-application version
;	of the debugger, and doesn't want a safe stack and direct
;	page to use instead of using those of the application that
;	was running at the time that the debugger was called.  Note
;	that the stack command only works if e = 0.
;
;	6)  Fixed bug that caused the stack pointer to be off by one
;	when returning from real time code in emulation mode via
;	a BRK instruction.]
;
; July 11, 1988	1)  Fixed bug in the loops that saved and restored the entry
;	state.
;
; July 12, 1988	1)  Fixed bug that caused the bank byte of the break vector
;	to be patched incorrectly in the rom version.
;
;	2)  Fixed bug that caused problems with real breakpoints
;
; July 13, 1988	1)  Fixed bug that caused real breakpoints to mess up on the
;	first J or X command executed after a trace or single
;	step with the display on.
;
; July 15, 1988	1)  If the option key is held down when the debugger is
;	first called during initialization, the debugger will
;	not be installed.
;
;	2)  Fixed bug that caused the debugger's memory not to be
;	deallocated by the QUIT command in the init version.
;
;	3)  Fixed bug that caused PC not to be set when entering
;	debugger via a break.
;
; July 18, 1988	1)  Moved allocation of 1k back to every time debugger is
;	run.
;
;	2)  Fixed bug that caused the PCK on entry not to be
;	restored on exit if in emulation mode.
;
;	3)  Debugger now recognizes CTRL-OA-Option-ESC while doing
;	real-time execution within the debugger.  This will not
;	re-enter the debugger recursively, but will just stop
;	execution of the code and drop you back to the main
;	debugger display screen.  When quitting, you will still
;	exit to the point at which the debugger was first called.
;
;	4)  The Prodos 16 stack based system call is now recognized,
;	and all JSLs to it are executed at real time.
;
;	5)  The PC was being set to two bytes past the break when
;	entering the debugger via a brk instruction.  Now it
;	is set to point to the brk instruction.
;
; July 19, 1988	1)  When doing the QUIT command, the debugger checks if the
;	user has patched the break vector since the debugger
;	patched it since initialization.  If not, then it is
;	restored to what it was before initialization.
;
;	2)  Version # is now followed by a letter indicating which
;	version of the debugger.  The letters are 'r' for the
;	rom version, 'i' for the init version, and 'a' for the
;	application version.
;
; July 21, 1988	1)  Changed user breaks to monitor breaks.  This just means
;	that the command to set breaks to drop into the monitor
;	is now 'mbrk' instead of 'ubrk.'  Also the flag that
;	indicates the type of break is now an 'm' instead of a 'u'.
;
;	2)  Changed commands around to remove almost all case
;	sensitivity.  To do this several things were changed.
;	First of all, the commands 'e=', 'm=', and 'x=' were
;	removed.  These can still be changed by toggling them with
;	the 'e', 'm', and 'x' commands, which must be in lower
;	case.  The old 'M=' and 'X=' commands now work in lower
;	case as well as upper case.  The old command 'D' which
;	was not case sensitive has been changed to 'DP:'.  This
;	is so that the old command 'd=' could be removed and
;	replaced with the new command 'd' which toggles the
;	disassembly mode.  The old command 'DP' was changed to
;	'DPAGE'.  This was done so as to avoid typing it
;	accidentally when meaning to type 'DP:'.  Finally, the
;	'quit' command was changed to 'unload'.
;
;	3)  Fixed bug that caused error when doing a command of the
;	form '1234X' or '1234J' from the dump screen.
;
;	4)  The command parser now only starts interpreting the 
;	currentcommand from the previous equal sign if that 
;	equal sign was from a valid expression.  This eliminates
;	some confusion from occurring when trying to set an 
;	invalid flag equal to a value.
;
;	5)  Fixed bug that caused numbers of more than 3 bytes that 
;	were entered on the command line to be truncated to 2 
;	bytes.
;
;	6)  The default base for numbers not preceded by any sign
;	in an expression is now hex.  Decimal numbers now must
;	alway be preceded by a positive or negative sign.
;	This means that if you want to add hexadecimal $1A to
;	decimal 5 for example, you must now type '1A++5='.
;
;	7)  The 'J' command (without an address in front of it)
;	was not working corectly.  The code was doing some strange
;	things that didn't seem to make much sense.  I added code
;	that should make it do the right thing.  Let's hope it
;	works OK.
;
;	8)  Made the letter 'G' interchangable with the letter 'J'
;	for jumping into code.  This was done because everyone
;	(myself) included thought that there already was a 'G'
;	command, but it didn't exist.
;
; July 22, 1988	1)  Changed the 'on' command so that it always causes the
;	master display screen to be displayed.  It used to ignore
;	the on command the screen was on, but we were viewing
;	a different screen by using the backslash commands.
;
; July 25, 1988	1)  Took out fix that stopped numbers of greater than 3 bytes
;	to be truncated.  It was causing too many problems.
;
;	2)  Fixed bug that caused CLOAD and CSAVE not to work in
;	ROM version.
;
;	3)  Fixed bug with the automatic switching back to the
;	master display screen when typing a command that
;	isn't available from the dump screen.
;
;	4)  Fixed bug that caused an expression of the form
;	'5A=' not to be converted to decimal.
;
; July 26, 1988	1)  Fixed bug that caused master display screen to be turned
;	back on automatically if doing a trace or single-step
;	while it was turned off.
;
; July 27, 1988	1)  Took out the letters following the version number.
;
; July 28, 1988	1)  Corrected the help screens.  Made changes so that there
;	are no invalid commands, but didn't add all the new
;	commands.
;
; Aug. 1, 1988	1)  Toolbox calls are now recognized during disassembly.
;	If a JSL $E10000 is seen, then the debugger will check
;	if the previous instruction was a LDX immediate and if
;	long indexing is on.  If both of these are true, then
;	the tool call name will be displayed instead of the
;	JSL E10000.  The LDX # will still be displayed.
;
;	2)  It is now possible to return from the monitor (after
;	entering it via the MON command) by pressing
;	CTRL-OPTION-OA-ESC.
;
; Aug. 3, 1988	1)  Fixed bug that caused the debugger to be messed up after
;	hitting a BRK or pressing CTRL-OA-OPTION-ESC while
;	tracing a subroutine in real-time.
;
;	2)  Fixed bug/feature that pushed an extra long return
;	address onto the stack when executing a JSR in real
;	time.
;
; Aug. 5, 1988	1)  Fixed another bug that caused the debugger to crash on
;	exit if a BRK was hit or CTRL-OPTION-OA-ESC was pressed
;	while inside a real time JSR.
;
;	2)  Debugger now restores the next instruction if a BRK was
;	encountered or if CTRL-OA-OPTION-ESC wass pressed while
;	tracing a subroutine in real time.
;
;	3)  The quit command now quits to the current PC and 
;	current state instead of to the point of entry.  The
;	screens are still set to what they were on entry.
;
; Aug. 10,1988	1)  Fixed bug that caused the high byte of both X and Y to
;	be destroyed when entering a subroutin in real time.
;
;	2)  Added an unload by user ID command.  This command is
;	of the form 'UNLOAD id#' and simply calls the
;	tool box routine DisposeAll with the  given ID #.
;
;	3)  The routine that checks for toolbox calls during
;	disassembly now no longer checks to make sure that
;	long indexing is on.  This was causing problems when
;	code was being entered with the mini-assembler.
;
;	4)  Added two new commands, 'TOOL #toolnum' and
;	'TOOL _ToolName'  Both of these commands output the
;	tool number and name of the specified tool.  If
;	an unknown tool number is requested, then the tool
;	name will be output as unknown.  If an unknown tool
;	name is requested, the debugger will just beep.
;
; Aug. 11, 1988	1)  Changed the UNLOAD command to SHUTDOWN.  It now takes
;	two arguments, the user ID and a quit flag, and executes
;	a UserShutDown c
all.  The command should be of the form
;	'SHUTDOWN id# quit_flag.'
;
;	2)  The TOOL command now also prints the actual address
;	of the tool routine.
;
;	3)  Pressing 'S' while in set mode now toggles the stack
;	subdisplay between showing stack addresses and stack
;	offsets.
;
;	4)  Pressing 'D' while in set mode now toggles the dump
;	alignment on and off.  If dump alignment is on, then
;	memory dumps will be aligned to 16-byte boundaries.
;
; Aug. 12, 1988	1)  Fixed bug that caused extra bytes to be left on the
;	stack when returning from a real time JSR.
;
;	2)  Fixed bug that caused major problems when entering real
;	time JSR's and JSL's in emulation mode.
;
;	3)  The debugger now recognizes GS/OS calls (JSL $E100A8)
;	and formats them in disassemblies.
;
; Aug. 18, 1988	1)  Added the ability to load and use templates.  A file of
;	templates can be loaded in with the command:
;	'LOADTEMP TemplateFile'.  An error $FFFF while loading
;	a template file means that the file is not in the correct
;	format. A template can be used with
;	the command '_TemplateName address' where address is
;	the address of your data.  A template file consists
;	of one or more templates.  Each template should start
;	with a line of the form '_START TemplateName' and end
;	with a line '_END'.  Anything not between a start and
;	end will be ignored.  Each line in the template should
;	be of the form 'Label type count'.  The fields should be
;	seperated by on or more tabs or spaces.  The count field
;	can be omited signifying just one value.  Types may
;	be either 'byte', 'word', 'long', 'ascii', 'cstring',
;	or 'pstring' and may be abreviated to their first letter.
;	All templates can be unloaded by the command 'KILLTEMP'.
;
; Feb. 2, 1989	1)  The user ID used for the stack and direct page segment
;	was being read of real zero page.  This random value sometimes
; 	worked and sometimes did not.
;
; Feb. 16, 1989	1)  Debugger now loads in a default configuration file on
;	startup.  This file should be called 'DEBUG.SETUP'.  It
;	should be located in the same folder as the debugger for
;	the init version, and should be located in the
;	SYSTEM/SYSTEM.SETUP folder of the bootup disk for the init
;	and ROM versions.
;
;	2)  If breakpoints are in when the BP command is entered, they
;	will automatically be taken out before the user edits the
;	breakpoints.  When the user is done editing the breakpoints,
;	they will automatically be put back in.
;
; Feb. 22, 1989	1)  Now check to see if the printer slot is selected before
;	allowing the screen to be printed.  Also make sure that slot
;	three is selected before allowing the MON command.  These
;	changes will supposedly prevent bugs in system 5.0.
;
; Feb. 22, 1989	1)  Fixed file so that tabs are set to 20.  It used to be
;	that everything was indented with spaces and some tabs.
;	This made it a real pain to edit the code.
;
; Feb. 23, 1989	1)  Added a scroll down (SCROLLDN) and a scroll dissasembly
;	subdisplay down routines.  These are to be used with viewing
;	history.
;
; Feb. 24, 1989	1)  Found and fixed three SEP #30's and two REP #30's.
;	These should have been SEP #$30 and REP #$30.  This fixed
;	the ten-speed bug and who knows what else.
;
; Mar. 3, 1989	1)  Changed quit command to QR (for quit resume) and put
;	back the old quit command as Q (or QUIT).  The Q command
;	restores the state of the machine to what it was on entry.
;
;	2)  Default config file now called GSBUG.SETUP not
;	DEBUG.SETUP.
;
;	3)  We no longer try to load a default config file in the
;	ROM version.
;
;	4)  No longer define ThisIsRAM in this file.  It must be
;	defined on the command line.
;
; Mar. 10, 1989	1)  Fixed bug that caused stack to be messed up when
;	tracing in emulation mode.  Now all of the save and
;	restore register routines are macros.  Before they
;	were copies of the same code.
;
;	2)  Debugger stack is now allocated as not locked
;	and is then locked afterwards.  This causes it to be
;	allocated near the top of bank 0, and not at $2000.
;	This fixes the bug that caused crashes when trying to
;	view page 2.
;
;	3)  Quit command no longer works from init or rom versions.
;	the RESUME command (R or RESUME) does what QR was doing.
;	UNLOAD unloads the debugger and then does a resume.  There
;	is no command that does what quit did.
;
;	4)  Added two new commands: SE (Save Environment) and
;	RE (Restore Environment).  SE saves a snapshot of the
;	current state of the registers.  RE will restore this
;	saved state.  If the RE command is used when no state
;	has been saved with the SE command, then the state of
;	the registers on entry will be restored.
;
; Mar. 16, 1989	1)  Fixed bug that caused set command to no longer work.
;
; Mar. 17, 1989	1)  Added saved trace history.  You can set trace save ranges
;	in the MP subdisplay.  The H command sets a window.  If
;	any windows are set, then only instructions inside of
;	the window are saved in the trace.  Open a trace history
;	file with the command 'OPENHIST filename'.  All instructions
;	in the trace history range will be saved to the file allong
;	with the current state of the registers.  Close the history
;	trace file with the command 'CLOSEHIST'.  View a history
;	trace with the command 'VIEWHIST filename'.  When viewing 
;	history, the current instruction is highlighted and the
;	registers reflect the state of the registers just before that
;	instruction was executed.  Use the up and down arrows to move
;	up and down one step.  Use the arrows with OA to move a
;	page at a time.  Use the left and right arrows to move to
;	the start or end.  Type a step number to move to that step.
;	Press esc to finish viewing the history.
;
;	2)  Added breaks on toolcalls and toolcall errors in trace mode.
;	To set a break on a toolcall use the command 'SetTBrk #toolnum'
;	or 'SetTBrk _toolname'.  To clear a break on a toolcall, use
;	the command 'ClrTBrk' followed by the name or number.  To clear
;	all breaks on a toolcall, use 'ClrAllTBrk'.  If a break is set,
;	execution will halt on the toolcall, before it is executed.  The
;	commands to set and clear breaks on tool errors are 'SetEBrk',
;	'ClrEBrk', and ClrAllEBrk'.  These will cause the trace to halt
;	if the given toolcall returns an error.  The debugger will halt
;	on the next instruction without executing it.  If either byte of
;	the tool call number equals zero, then that byte will match any
;	high byte.  For example, setting a break on $0004 will cause a
;	break on any Quick Draw call, while $0000 will cause a break on
;	any call.  Up to twenty of each kind of break can be set.  The
;	set breaks will be saved with your configuration.  To view the
;	currently set breaks, use the command 'ShowBrks'.
;
;	1)  Fixed expresions, hex #'s weren't right.
;
;	2)  Fixed check for $C017 before entry into mon.
;
; March 20, 1989	1)  Can now break on tool calls and tool call errors while
;	in real time.  This can be done by using the commands
;	TBRKIN and EBRKIN to set real time tool breaks and real time
;	tool error breaks respectively.  To disable real time
;	breaks, use the commands TBRKOUT and EBRKOUT.
;
;	2)  Fixed problem that caused the PC to incorectly reflect
;	the current instruction when halting from a trace while the
;	display is off.
;