« first day (2854 days earlier)      last day (64 days later) » 

LdBeth
LdBeth
15:24
@Adám There are several related articles on jsoftware.com jsoftware.com/papers/AFIPS196212.htm jsoftware.com/papers/PNSD.htm
LdBeth
LdBeth
15:38
And a citation search on googles scholar with keyword "Intel" "8086" are either not related to APL or not actually on micro programming, except for one that only briefly mentioned Intel 8008 dl.acm.org/doi/abs/10.1145/800203.806252
Adám
Adám
@LdBeth Right, IN was not used for any later systems like Intel's.
 
2 hours later…
drab
drab
18:06
Here is a bug in at least version 18 and 19.

It is an incorrect domain error trying to get unique rows of a simple 2 by 2 numeric matrix

∪ 2 2⍴ 2 1E¯323 2 1.797693134862316E308

The problem has something to do with the extreme range of the values.
Adám
Adám
@drab Oh no! Thank you for reporting. Will log that right away.
Adám
Adám
18:18
(Logged as issue 21742)
 
1 hour later…
drab
drab
19:28
Here is another bug.

Version 19 gives a wrong answer for a binomial involving a very large integer

9007199254740992 ! ¯1
1.603810890548638E¯28

The correct answer is 1 since the integer is even.

That big integer is the largest exact integer stored as a floating point.

Version 18 gives a domain error for this same example, which is better than a wrong answer.
Adám
Adám
20:10
@drab Wrong value is worse than error. Logged as 21743. Thank you so much!
Silas
Silas
20:21
imagine a precision error - with ⎕fr←1287, 1=9007199254740992 ! ¯1
 
3 hours later…
David
David
23:36
Does anyone know where one can find online the late 1960s IBM Systems Journal article on APL used to rigorously define the IBM System/360 instruction set?

Maybe Adam knows....

« first day (2854 days earlier)      last day (64 days later) »