SCOUG-SundialSIG Mailing List Archives

<< Previous Message << >> Next Message >>

Date:	Thu, 31 May 2001 09:44:18 PDT
From:	Peter Skye <pskye@peterskye.com >
Reply-To:	scoug-sundialsig@scoug.com
To:	scoug-sundialsig@scoug.com
Subject:	SCOUG-SundialSIG: What, if anything, happened?

Content Type: text/plain

Lynn H. Maxson wrote:
>
> you could have solved this using PL/I

Lynn knows my weak spot -- PL/I is my all-time favorite language.

> and declaring the field to be "fixed bin (31,5),
> a fullword binary number with a fractional part.

There were two considerations:

-- First, ease of entering the data via the keyboard. By eliminating
the "/", there are fewer keystrokes and fewer chances for entry error.

-- Second, my software was originally developed for 286 and 386
machines. Lynn is correct that a fullword binary number will work, and
in fact that's what I used. The only difference is that I used a
"scaling factor" on the data so there was no fractional part. When the
data was read into memory the input routine scanned it for the
fractional parts that were used. If there were, for example, 8ths but
no 16ths then the scaling factor was 8 and every data value was
multiplied by 8 (2-1/8 would become 17). Once processing was done the
answers were scaled back to the "proper" values. By using fullword
binary numbers (32 bits including sign bit) instead of floating point,
the calculations ran much faster. (My early compiles used 16 bit
values, which gave a typical price range of 1/16 to $2000. I wrote my
own LIM access routines to get around memory thrashing.)

> Of course, if you are using REXX, C, PASCAL,
> COBOL, FORTRAN, JAVA, C++, LISP, PROLOG,
> or even assembler, you can't do that.

Why not? I used assembler and Pascal. Do you mean the (31,5) part with
the compiler tracking the fractional point during calculations? Yes,
then you're correct that "you can't do that" -- although the execution
speed is the same. At the time I didn't have a PL/I compiler because I
was running DOS (my Digital Research PL/I compiler ran on CPM). I now
have IBM's PL/I.

> Just note that this capability, which has never
> been reproduced in any other programming
> language, has been available since 1964.

Lynn is referring to the original PL/I language specification, developed
in 1964. Lynn, didn't they swipe that from Algol?

> Peter has provided excellent insight into how
> he processes, stores, maintains, and retrieves
> his financial data. May I suggest that he
> expand it into a more formal presentation
> combined with a Q&A to serve as the topic
> for our next Sundial SIG meeting in June?

I might be over at the Programming SIG trying to run EMX compiles with
Dallas and Steven.

Just store the data in the format you receive it in, in any filename you
like. Let the program figure out and "learn" where the data is.

- Peter

=====================================================

To unsubscribe from this list, send an email message
to "steward@scoug.com". In the body of the message,
put the command "unsubscribe scoug-sundialsig".

For problems, contact the list owner at
"rollin@scoug.com".

=====================================================

Return to [ 31 | May | 2001 ]

SCOUG, Warp Expo West, and Warpfest are trademarks of the Southern California OS/2 User Group. OS/2, Workplace Shell, and IBM are registered trademarks of International Business Machines Corporation. All other trademarks remain the property of their respective owners.