Translation of TBStrLib functions to a generic Ada95 package

cjiii

Before undertaking my translation of 20-string functions in TB into Ada95, for the portable embedding of my theorem prover M8/VL4, I thought to ask if this Ada code already exists.

Thank you in advance for responses off-line to info@esatz-systems.com or info@cec-services.com .

Colin James III
Researcher affiliated with WTAMU
Colorado Springs, CO
(719) 210-9534

cjiii

Since many of the TB string functions use the CALL Divide (w, x, y, z) subroutine with two exceptions of overflow and division by zero, is the source code for that available?

mcc

Hello Colin,

The source for that SUB is found in the
TB Silver Manual, under Statements, Built-in Functions and Subroutines
pp. 237-8:

DIVIDE Subroutine
CALL DIVIDE (numex, numex, numvar, numvar)
CALL DIVIDE (dvend, dvsor, q, r) divides dvend by dvsor to give quotient q and remainder r. More
specifically, q and r are solutions to dvend = dvsor*q + r, where q = INT(dvend/dvsor) and r = MOD
(dvend, dvsor). For example:

! DIVIDE with these arguments ! Returns
CALL DIVIDE (2.5, 1.5, q, r) ! q = 1, r = 1
CALL DIVIDE (1, 10, q, r) ! q = 0, r = 1
CALL DIVIDE (7, 3, q, e) ! q = 2, r = 1
CALL DIVIDE (-7, 3, q, r) ! q = -3, r= 2
CALL DIVIDE (7, -3, q, r) ! q = -3, r = -2
CALL DIVIDE (-7, -3, q, r) ! q = 2, r = -1
Exceptions: 1002 Overflow.
3001 Division by zero.

Regards,
Mike C.

cjiii

Thanks Mike.
I was not clear that I wanted the full TB source code for the routine, including the exception handling wrapper, in other words the complete source code looking like the functions in StrLIB. It's easier to translate into Ada95 that way rather than cooking up the code in TB from the description and putting a wrapper around it: anything to reuse code!

cjiii

Here is my rendition of the resuscitation of the Div routine ....

!
! -- Div
!
! -- DIVIDE Subroutine
! -- CALL DIVIDE (numex, numex, numvar, numvar)
! -- CALL DIVIDE (dvend, dvsor, q, r) divides dvend by dvsor to give quotient q and remainder r. More
! -- specifically, q and r are solutions to dvend = dvsor*q + r, where q = INT(dvend/dvsor) and r = MOD
! -- (dvend, dvsor). For example:
! -- Statements, Built-in Functions and Subroutines 237
! -- ! DIVIDE with these arguments ! Returns
! -- CALL DIVIDE (2.5, 1.5, q, r) ! q = 1, r = 1
! -- CALL DIVIDE (1, 10, q, r) ! q = 0, r = 1
! -- CALL DIVIDE (7, 3, q, e) ! q = 2, r = 1
! -- CALL DIVIDE (-7, 3, q, r) ! q = -3, r = 2
! -- CALL DIVIDE (7, -3, q, r) ! q = -3, r = -2
! -- CALL DIVIDE (-7, -3, q, r) ! q = 2, r = -1
! -- Exceptions: 1002 Overflow; 3001 Division by zero.

!
! -- Div as routine
!
! -- LET max_num = MAXNUM
! -- PRINT
! -- PRINT "MAXNUM, -1 * MAXUM):"; INT( max_num); INT( -1 * max_num)
! -- PRINT
! -- GET KEY zzz
!
! Copyright notice
!
PRINT
PRINT TAB( 5);
PRINT "TrueBASIC source code resuscitation of the Div routine for translation to Ada95"
PRINT
PRINT TAB( 5); "© Copyright 2024 by Colin James III All rights reserved."
PRINT
PRINT
DO
READ IF MISSING THEN EXIT DO: dividend, divisor
LET quotient = 0
LET remainder = 0

CALL Div( dividend, divisor, quotient, remainder)
PRINT "dividend, divisor, quotient, remainder ";
PRINT dividend, divisor, quotient, remainder
LOOP

PRINT
PRINT "Routine tested correctly on "; Date$; "."
PRINT
PRINT "Press any keys to end: ";
GET KEY zzz
PRINT

!
! Test data
!
DATA 2.5, 1.5 ! q, r = 1, 1
DATA 1, 10 ! q, r = 0, 1
DATA 7, 3 ! q, r = 2, 1
DATA -7, 3 ! q, r = -3, 2
DATA 7, -3 ! q, r = -3, -2
DATA -7, -3 ! q, r = 2, -1
DATA 1, 0 ! 3001 Division by zero

SUB DIV( dividend, divisor, quotient, remainder)
WHEN EXCEPTION IN
LET quotient = INT( dividend / divisor)
LET remainder = MOD( dividend, divisor)
USE
IF divisor = 0 THEN
PRINT
PRINT "Exception raised below."
PRINT
PRINT "3001 Division by zero"
ELSE
!
! The only other possible exception is an integer number greater than MAXNUM.
!
PRINT
PRINT "Exception raised below."
PRINT
PRINT "1002 Overflow"
END IF
END WHEN
END SUB ! DIV( dividend, divisor, quotient, remainder)

END ! Div as routine

mcc

That DIVIDE routine is one of the 259 routines and statements that are BUILT INTO TB. There IS no source code available for any of the TB internal code. They no longer have the fundamental source code that Kemeny and Kurtz wrote at the very beginning. That's the sad reason TB hasn't been and can't be "Upgraded" ( e.g., to be able to recognize a mouse wheel which didn't exist back then.).

So what I gave you is the only description of what "DIVIDE" is/does; i.e., that which is given in the manual. The library sources such as found in strlib, etc, are available because they were written afterwards and provided as extras.; they were never a part of the original "core" code. But the internal "source" code of the TB "core" has never been availble. Possibly it was way back when it was first invented at Dartmouth, and when it was used on a time-sharing teletype setup back in the '60's, and when HP adopted it for use in their 9800 series machines. Microsoft and others used it in watered-down versions., all of which came out as "PC"s became available in the early 80's.

I think there are a few scraps of the original Dartmouth Basic assembly or machine code (which is what TB actually is) still floating around somewhere, but apparently there is no one who has it who can figure it out. The only thing left is the "TB core" that you get when you buy any of the various TB versions. They all have the exact same "innards". Silver has essentially everything True Basic itself can natively do, and Gold merely adds a few extra things like being able to add dll's if you can code the dll in in C++. Bronze lacks the ability to bind to an exe. But the internal code with all the "built-in" stuff is is the same for all three and there is no source code available.

Regards,
Mike C.

cjiii

Please contact off-line for more info and pricing at info@cec-services.com .
Lessons learned are legion, constituting an Ada95 conference academic paper. It was rejected due to criticism of Ada as designed by programmers who were not mathematicians (same for C), unlike TB as designed by mathematicians for programmers.
Here is an example of the gymnastics performed to implement the TB Round( x, n) as mathematically correct into Ada95:

FUNCTION Round( x_flt: Float ; n_int: Integer) RETURN Float IS
-- The ** means exponent in Ada
-- It is a mistake here to assign the ( 10.0 ** n_int) to a Float variable because
-- that auto-rounds the expression, potentially to lose precision.
-- Hence the literal is repeated twice below to enforce precision.
-- The 10.0 forces a float, and the sign of n_int gives correct decimal results.
BEGIN -- The 'ROUNDING is an attribute in Ada for rounding up.
RETURN Float'ROUNDING ( x_flt * ( 10.0 ** n_int)) / ( 10 ** n_int) ;
END Round ;

This replicates the following TB examples for exponents -3 to 2:
Round( 123.4567,-3) to produce expected 123.457
Round( 123.4567,-2) to produce expected 123.46
Round( 123.4567,-1) to produce expected 123.5
Round( 123.4567, 0) to produce expected 123
Round( 123.4567, 1) to produce expected 120
Round( 123.4567, 2) to produce expected 100