OK, I think I get it, thanks. The only final point I would make is that the 
necessity to "clean up your own garbage" should be
a little more explicit in the J Wiki / J website. I would add something myself 
to the Wiki, but I don't have access and there seems
to be no section in the Wiki regarding OOP.
e.g. google search 
object orientation site:jsoftware.com
gives JForC and Learning J explanations of OOP, but not much else.
Then again, maybe this was clear to everyone else and it was just my  reading 
comprehension that was lacking when reading JforC/Learning J's
explanation of OOP (they both mention codestroy, but I didn't put 2 and 2 
together.)


--------------------------------------------
On Fri, 10/13/17, Raul Miller <[email protected]> wrote:

 Subject: Re: [Jprogramming] Garbage Collection and Objects
 To: "Programming forum" <[email protected]>
 Date: Friday, October 13, 2017, 12:26 AM
 
 Objects are, themselves, named
 collections of named values.
 
 It's not just that they are referenced via
 box literals - they are
 just a different
 flavor of classes, and are not GC'd because they use
 that same design. You can always reference
 objects via conl 1.
 
 This
 design does, unfortunately, make it much harder to
 reference
 unallocated memory. But the cost
 is that you can inspect arbitrary
 objects
 from the J command line.
 
 Thanks,
 
 -- 
 Raul
 
 On Thu,
 Oct 12, 2017 at 10:56 AM, 'Jon Hough' via
 Programming
 <[email protected]>
 wrote:
 > By "explicitly
 release" you mean dereference in the following sense
 right?
 > thingy=:  ? 1000 * 1000 $ 0 NB.
 Lots of memory allocated.
 > thingy=. 1
 NB. The 1000x1000 matrix was dereferenced and is eventually
 GCed.
 >
 > But objects
 don't work like this:
 > thingy =: 100
 conew 'A' NB. Memory allocated for instance of A
 > thingy=: 1 NB. instance of A was
 dereferenced, but memory not reclaimed - memory leak.
 >
 > My only question is,
 why not automatically reclaim dereferenced objects?
 > I guess it is difficult to decide if an
 object is no longer reachable, since objects
 > are referenced as boxed literals. Any
 boxed literal could inadvertently "reference"
 > an object that would otherwise be ready
 for GC.
 > In an imaginary J with a
 "Pointer" datatype, or whatnot, it would be
 > easier to figure out that an object is not
 reachable, just by reference counting
 >
 the "Pointers" to the object.
 >
 >
 --------------------------------------------
 > On Thu, 10/12/17, Raul Miller <[email protected]>
 wrote:
 >
 >  Subject:
 Re: [Jprogramming] Garbage Collection and Objects
 >  To: "Programming forum" <[email protected]>
 >  Date: Thursday, October 12, 2017, 9:21
 PM
 >
 >  Anything
 involving names in
 >  locales needs
 explicit GC.
 >
 >  So,
 for example:
 >
 >    
 thingy=: i.1e6
 >
 > 
 That name - thingy - will persist until you
 >  explicitly release it.
 >
 >  Locales actually
 make this a bit easier (since
 >  you can
 erase the
 >  entire locale rather
 than
 >  having to erase all the names in
 it).
 >
 >  The problem
 comes when you have been trained to
 > 
 use millions of tiny
 >  objects.
 That's a
 >  bad habit not just
 because of "GC", but because
 >  it's
 >  a poor
 use of space and time, in J.
 > 
 Generally speaking, in J you want
 >  to
 arrange
 >  things so the low level
 structures are regular and the
 > 
 complexity bubbles up to the top levels. This
 >  tends to be great for
 >  comprehension, but
 >  painful or worse when you are not
 supposed to be
 >  able to understand
 what you are working on.
 >
 >  Thanks,
 >
 >  --
 >  Raul
 >
 >  On Thu, Oct 12,
 2017 at 5:50 AM, 'Jon
 >  Hough'
 via Programming
 >  <[email protected]>
 >  wrote:
 >  > Is
 there a reason J doesn't
 >  perform
 GC on objects? I was unaware we had to destroy our
 >  own objects
 >  >
 (In retrospect, I guess the
 >  existence
 of codestroy was a hint ).
 >  >
 >  > Example:
 > 
 >
 >  > oclass 'B'
 >  >
 >  >
 create=: 3 : 0
 >  >
 >  > Mat=: ? (y,y) $ 0
 >  >
 >  )
 >  >
 >  >
 destroy=:
 >  codestroy
 >  >
 >  >
 >  >
 >  >
 >  >
 >  coclass
 'A'
 >  >
 >  >
 >  >
 create=: 3 : 0
 >  > iterations=: y
 >  > myB=:
 > 
 ''
 >  > )
 >  >
 >  >
 >  > runLoop=: 3 : 0
 >  > ctr=: 0
 >  >
 while. ctr
 >  < iterations do.
 >  > myB=: 400 conew
 >  'B'
 >  >
 ctr=:>:ctr
 >  > end.
 >  >
 > 
 'finished'
 >  > )
 >  >
 >  >
 >  >
 >  destroy=:
 codestroy
 >  >
 > 
 >
 >  > myA=: 1000 conew
 >  'A'
 >  >
 runLoop__myA 0
 >  > NB. Let myB
 reference an int now. It might
 >  be
 expected that the
 >  > NB. 400x400
 >  matrix's memory allocation was
 freed, but it
 >  > NB. is still
 there.
 >  >
 > 
 myB__myA=: 1
 >  >
 >  >
 >  Viewing
 memory usage in htop or Activity Monitor, this
 >  program goes into the Gigabytes quickly
 and as far as I can
 >  see the
 >  > memory is never reclaimed.
 >  > I am not complaining, I am just
 wondering,
 >  why unreferenced objects 
 are not GCed,
 >  > and also recommend
 that OOP explanations
 >  in the Wiki,
 JforC (assuming a new edition)  be a little
 >  more explicit
 > 
 > in the necessity of
 > 
 codestroy.
 >  >
 >  > This is
 > 
 A's runLoop that destroy unreferenced objects
 >  > runLoop=: 3 : 0
 >  > ctr=:
 >  0
 >  > while. ctr < iterations do.
 >  > if.-. myB -: '' do.
 >  > destroy__myB ''
 >  > end.
 >  >
 myB=: 400 conew
 >  'B'
 >  > ctr=:>:ctr
 >  > end.
 >  >
 >  'finished'
 >  > )
 >  >
 >  >
 >  >
 >  Using this, memory allocation does not
 increase. I'm
 >  sure this is
 probably obvious to J experts, but as far as I
 >  can see, doing there is
 >  > no explicit
 > 
 explanaiton of this anywhere.
 >  >
 > 
 ----------------------------------------------------------------------
 >  > 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

Reply via email to