I'll look into this.
Henry Rich
On 8/16/2016 7:38 AM, bill lam wrote:
Apologize to re-opening this old thread. I tried Joe's
patch on the current J source and it seems working. But
it also crashed when running g640 in the test suite.
I can get a simple repeatable expression that caused crash.
?@:(5e3&$)"0 ] 2
but 1e3 or 2e3 is ok. Any idea?
Чт, 17 дек 2015, Joe Bogner написал(а):
looks like the problem likes in
https://github.com/openj/core/blob/18fd23bbdc2f50770eb3047e978cd5e4e3b47039/vrand.c#L726
It looks like it checks to see if the first number in the list is 2
and if so, it uses the roll2 implementation
RZ(w=vi(w)); m=*AV(w);
if( 2==m)RZ(z=roll2 (w,&b));
It also appears to be a 64-bit bug
32-bit:
?. 16 # 2 2 2 16
1 0 0 1 0 1 0 0 0 0 1 0 1 0 0 0 0 0 1 0 0 0 0 1 1 1 0 0 1 1 0 0 0 1 1
1 1 1 1 0 1 1 1 0 0 1 1 1 13 12 13 4 9 15 5 9 6 3 9 1 2 11 1 8
64-bit
?. 16 # 2 2 2 16
0 1 0 1 1 0 0 1 1 1 0 0 0 0 1 0 0 1 0 1 1 0 0 0 0 0 1 0 0 1 0 1 0 0 0
0 0 0 1 0 1 1 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 1 1 1 0 1 1
The simplest 'fix' would be to disable the roll 2 optimization by
commenting it out, but that may not be ideal since presumably it was
put in place for a reason to speed up random booleans on large arrays.
The next fix would be to check to see if all numbers in the ravel are
2 and only then use roll2. The last, probably more complicated fix
would be to support 2 and other numbers in roll2 safely in 64-bit, but
even that is pretty simple
just copy and paste the block of shifts (since there is only 8), to
add 8 more for 64-bit... need to do this for the random and fixed seed
part
}else for(j=0;j<q;++j){
t=NEXT;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t;
//start new code
#if SY_64
t>>=1;
if(!(2==*v++&&2==*v++&&2==*v++&&2==*v++))R mark; *zv++=mk&t; t>>=1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark;
*zv++ = mk&t; t >>= 1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark;
*zv++ = mk&t; t >>= 1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark;
*zv++ = mk&t; t >>= 1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark;
*zv++ = mk&t; t >>= 1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark;
*zv++ = mk&t; t >>= 1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark;
*zv++ = mk&t; t >>= 1;
if (!(2 == *v++ && 2 == *v++ && 2 == *v++ && 2 == *v++))R mark; *zv++ = mk&t;
#endif
}
On Thu, Dec 17, 2015 at 10:29 PM, Linda A Alvord
<[email protected]> wrote:
These are the results you should get in a fresh version of JHS or jqt. If
you use ?. you will insure repeated exact results in a session. In all cases
the length of the list is 64 elements so there should be enough 2's.
?. 16 # 2 2 2 16
1 0 0 1 0 1 0 0 0 0 1 0 1 0 0 0 0 0 1 0 0 0 0 1 1 1 0 0 1 1 0 0 0 1 1 1 1 1
1 0 1 1 1 0 0 1 1 1 13 12 13 4 9 15 5 9 6 3 9 1 2 11 1 8
?. 16 # 3 2 2 16
0 2 2 1 2 0 2 1 1 2 0 2 1 1 1 2 0 0 0 0 1 0 0 0 0 1 1 1 0 0 1 1 0 0 0 1 1 1
1 1 1 0 1 1 1 0 0 1 13 11 13 12 13 4 9 15 5 9 6 3 9 1 2 11
?. 16 # 16 2 2 2
2 11 15 4 6 7 4 1 4 2 6 14 1 14 7 12 0 0 0 0 1 0 0 0 0 1 1 1 0 0 1 1 0 0 0 1
1 1 1 1 1 0 1 1 1 0 0 1 1 1 1 0 1 0 1 1 1 1 0 1 1 1 0 1
$?. 16 # 2 2 2 16
64
$?. 16 # 3 2 2 16
64
$?. 16 # 16 2 2 2
64
Linda
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of 'Pascal
Jasmin' via Programming
Sent: Thursday, December 17, 2015 12:41 PM
To: [email protected]
Subject: Re: [Jprogramming] bug in ?
?. 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16
0 0 0 0 1 1 0 0 0 1 1 1 0 0 0 0 1 0 0 0 0 1 1 1 1 1 1 0 1 1 1 0 1 0 1 1 1 1
0 1 1 1 1 0 0 1 1 1 8 9 6 8 14 7 12 9 1 15 2 15 3 13 3
?. 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16
0 1 0 1 1 0 0 1 1 1 0 0 0 0 1 0 0 1 0 1 1 0 0 0 0 0 1 0 0 1 0 1 0 0 0 0 0 0
1 0 1 1 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 1 1 1 0 1 1
one less leading 2 in first example. 47 leading 2s, correct result. 48
leading 2s and everything is 2.
----- Original Message -----
From: Raul Miller <[email protected]>
To: Programming forum <[email protected]>
Sent: Thursday, December 17, 2015 12:28 PM
Subject: Re: [Jprogramming] bug in ?
Consider
?.2 2 2 32
0 1 0 18
Thanks,
--
Raul
On Thu, Dec 17, 2015 at 12:13 PM, 'Pascal Jasmin' via Programming
<[email protected]> wrote:
?. 16 # 2 2 2 16
result is same as
?. 16 # 2 2 2 2
ignores the request for 0-15 range random numbers.
----- Original Message -----
From: Raul Miller <[email protected]>
To: Programming forum <[email protected]>
Sent: Thursday, December 17, 2015 12:10 PM
Subject: Re: [Jprogramming] bug in ?
On Thu, Dec 17, 2015 at 11:49 AM, 'Pascal Jasmin' via Programming
<[email protected]> wrote:
?. 16 # 2 2 2 16
0 1 0 1 1 0 0 1 1 1 0 0 0 0 1 0 0 1 0 1 1 0 0 0 0 0 1 0 0 1 0 1 0 0 0 0 0
0 1 0 1 1 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 1 1 1 0 1 1
?. 16 # 3 2 2 16
0 2 1 0 2 1 1 1 2 1 0 2 1 2 1 2 1 0 1 1 1 1 1 1 0 0 1 0 0 1 1 0 0 1 1 1 0
1 0 0 0 1 1 1 1 0 1 0 8 4 14 3 1 6 14 10 15 13 5 6 2 14 0 5
The last 16 numbers in the first example should match the 2nd.
Why do you say that?
Thanks,
--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
