-------------------
				ACEcalc version 2.0
				------------------- 

					by

			    David J. Benn, November 1995


CONTENTS
--------
° General Description
° Order of Operations
° Accuracy of Calculations
° Error Messages
° Inside the Program
° Problems/Limitations
° Changes
° Contacting the Author
            

General Description
-------------------
ACEcalc is Workbench calculator which allows for the construction and
evaluation of fixed-point infix expressions.

Legal operators are: +,-,*,/,^ corresponding to addition, subtraction,
multiplication, division, exponentiation. Two additional operators: (
and ) allow sub-expressions to be parenthesised.

Unary functions currently provided are: SIN, COS, TAN, LOG, EXP, SQR,
and INT which correspond to sine, cosine, tangent (these expect radians),
natural logarithm, e raised to the power of n, square root and integer
conversion with rounding. All functions take a single argument.

Expressions are entered either by clicking buttons on the calculator 
with the mouse, or by using the Amiga's keyboard. In the latter case,
the return/enter key has the same effect as "=", and the destructive
backspace and Del keys correspond to the "«-" button on the calculator.

Note that all buttons in the first four rows have keyboard equivalents,
but that none in the last two rows do.
 
There are three other calculator buttons of note: CLR, STO, RCL. The
first clears the calculator's display. STO stores the current expression
(shown in the calculator's display). RCL retrieves the most recently
stored expression and displays it. There is only a single memory. In
order to clear the memory, select CLR and STO in sequence.

Typical ACEcalc usage is somthing like this (comments in curly braces):

	(12+2)*3.5		{enter this expression}
	=			{"=" button or keyboard equivalent}
	49			{result appears in display}
	STO			{store 49}
	CLR			{clear the display}
	SIN(			{enter "SIN("}
	RCL			{recall memory contents}
	)			{closing parenthesis}
	=			{"=" button or keyboard equivalent}
	-0.9537535		{result of SIN(49) appears in display}
	
The project menu has items for an About box and quitting. The program
may be exited by selecting the latter from the Project menu, hitting
Amiga-Q or clicking the calculator's close button.

Note that the calculator window may be dragged around the screen and
depth arranged.


Order of Operations
-------------------
Expression evaluation proceeds from left to right, with operation
precedence as follows:

	1. 	()	
	2. 	SIN,COS,TAN,LOG,EXP,SQR,INT
	3.	^
	4.	*,/
	5.	+,-


Accuracy of Calculations
------------------------
Since ACEcalc only uses single-precision floating point math, the accuracy
of results is limited to around 8 integer digits or 7 decimal places, and 
beyond that, digits should be treated with suspicion.

A big advantage of single-precision math (Motorola FFP in this case) is
that it's fast. However, when ACE (see "Inside the program") supports 
double-precision (IEEE) floating point math, so will ACEcalc.

   
Error Messages
--------------
If an expression is malformed and cannot meaningfully be evaluated,
ACEcalc displays the message: SYNTAX ERROR.

If a divisor evaluates to zero the DIVISION BY ZERO message is displayed.

You should never see the STACK OVERFLOW or STACK UNDERFLOW messages. Let 
me know if you do.


Inside the Program
------------------
ACEcalc was written in ACE BASIC, a freeware AmigaBASIC compiler (developed 
by yours truly).

The calculator relies heavily upon ACE's GUI capabilities for the
creation and handling of menus, buttons etc.

The expression evaluation engine is a recursive descent parser.


Problems/Limitations
--------------------
° While ACEcalc displays exponential format numbers when a result is
  too large to be displayed in fixed point format, such results can
  not be used as components in subsequent calculations.


Changes
-------
Version 2.0	- Buttons now use topaz 8 font to avoid imagery mangling
		  when other default screen fonts are used.

		- Bug in exponentiation fixed: use of negative exponents
	  	  resulted in incorrect results. Also, since ACE v2.39
		  (unreleased at this time) uses single-precision IEEE
		  if available on the system in use, the results of
		  exponentiation are in general more accurate.

		- MsgBox function usage replaced by MsgBox statement.
  
    
Contacting the Author
---------------------
You can reach me in a number of ways:

	Internet:	D.Benn@appcomp.utas.edu.au
	
	CompuServe:	100033,605

	Phone:		(003) 261 461 [AH]

	Snail Mail:	9 Mayne Street,
			Invermay, Tasmania,
			Australia, 7248


				(*---oOo---*)