I have found the source of the memory leak & have to verify the fix.

Henry Rich

On 5/19/2017 7:15 PM, Jose Mario Quintana wrote:
Further testing suggests that the memory leak bug,

]^:0: (;/i.1e6)

is pervasive in the following sense:

It occurs when the verb argument is a boxed list with several items and for
most (any?) verbs  u  and  v  in the monadic or dyadic form  u^:v  ; it
affects J805 and J806, all flavors (JQt, JHS, jconsole), and platforms
(Windows, Linux, etc.).

Beware: if another verb contains u^:v  and it is executed repeatedly memory
is consumed and it can potentially freeze a computer; that is what happened
when I was bitten for the first time (I had to pull the plug).

PS.  The wicked crash also seems to be widespread.


On Tue, May 16, 2017 at 7:05 PM, Jose Mario Quintana <
[email protected]> wrote:

Thank you

On Tue, May 16, 2017 at 6:48 PM, David Mitchell <[email protected]>
wrote:

It ran in J64-804


On 5/16/2017 18:09, Jose Mario Quintana wrote:

0. Wicked Crash in JQt805/806

     Cloak=. ((5!:1)@:<'Cloak')Cloak=. (0:`)(,^:)
     'evoke fix amper'=. <@:Cloak "0@:;: '`:  f. &'
     train=. evoke&6 f.
     an=. <@:((,'0') ,&< ])

     indet=. '' ($ ,) 128!:5  NB. Is Indeterminate?
     eval=. {. train@:, an@:((train :: ])&.>)@:}.
     body=. ([ amper >@:{.)@:([ , <@:])`(eval@:}.@:[)@.(indet @:])
     GENOME=. <@:fix'body'

     vs=. GENOME&body f.
     S=. vs ]`''  NB. Strand notation (_. is used as a guard)

     S 'this' 'is' 'strand' 'notation' 0 a: (a.{~97+i.10) _.
┌────┬──┬──────┬────────┬─┬──┬──────────┐
│this│is│strand│notation│0│┌┐│abcdefghij│
│    │  │      │        │ ││││          │
│    │  │      │        │ │└┘│          │
└────┴──┴──────┴────────┴─┴──┴──────────┘

     S (body f.)`'' _. NB. Nothing is produced...
     S  NB. ... J805/806 Crashes!

     It runs fine in a custom J (Jx804) based on J804.  Can anybody
confirm
that it runs in J04 (I do not have one handy)?

     assert (,GENOME) -: S (body f.)`'' _.

     The problem seems to be related to in-place operations; but, I have
no
idea if it only can occur in wicked contexts or not.


1. Memory leak in JQt805/806

     This is one hurdle preventing us from moving an application from
Jx804
to Jx805/Jx806,

     ]^:0: ;/i.1e6
----------------------------------------------------------------------
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


---
This email has been checked for viruses by AVG.
http://www.avg.com

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to