RE: Bug in number conversion (?)

     Perhaps you are running afoul of the legacy number conversion issue.

     The legacy from PolyFORTH is to take numbers (ascii strings convertable
to numbers) with punctuation like 3:30:47 and 4/14/00 and 86,400 and 3.3 as
double numbers. It came in handy when stack items where 16 bits long and
86,400 seconds in a day would not fit in an unsigned single number. It has
come up before on this mailing list.

     If I may say so without hurting anyone at Forth, Inc.'s feelings it
might be time to retire this feature. Free advice, perhaps worth what you
paid for it.

     It could be worse. There are folks at Raytheon who expressed time of
day as a floating point number. Perhaps because it wouldn't fit in a FOTRAN
integer. The round off error accumulates and moves the RADAR range gate and
the missile misses. 20% in 8 hours.

http://www.cs.unc.edu/~dm/UNC/COMP205/LECTURES/ERROR/lec23/node4.html

> ----------
> From: Jeffrey R. Fox[SMTP:Jeffrey.Fox_at_wcom.com]
> Reply To: sftalk_at_forth.com
> Sent: Thursday, June 29, 2000 1:28 PM
> To: sftalk_at_forth.com
> Subject: [sftalk] Bug in number conversion (?)
>
> I recently typo'd in "3+" in a definition and found that I got the double
> number
> 3.0 -- that is surely not the intended behavior of the number conversion
> routines, is it?
>
> Jeff
>
> _______________________________________________
> Sftalk mailing list Sftalk_at_forth.com
> Visit Sftalk on the web at http://www.forthinc.com/mailman/listinfo/sftalk
>
Received on Thu Jun 29 2000 - 14:46:09 PDT