From: Abder-Rahman Ali on
Thanks a lot James for your advice. Yes, think have to read more deeply
about garbage collection.
--
Posted via http://www.ruby-forum.com/.

From: andrew mcelroy on
On Tue, Jul 13, 2010 at 11:24 AM, Abder-Rahman Ali
<abder.rahman.ali(a)gmail.com> wrote:
>
> Thanks a lot James for your advice. Yes, think have to read more deeply
> about garbage collection.

This looks like the perfect opportunity to send this conversation way
into technical land.

Consider the following presentation and then GC system:
Joe Damato does a great job explaining GC under Ruby.
http://timetobleed.com/garbage-collection-and-the-ruby-heap-from-railsconf/

I have also been reading up on a GC system called Schism- a real-time,
concurrent  non-moving, non-copying GC for java. I contacted the
university listed as being a collaborate on this Schism. Sadly it is
proprietary.
 However, if I could ever free up some time, I was going to check how
patent encumbered it is. If the road is clear, then this would be a
great thing to have in ruby.
www.filpizlo.com/slides/pizlo-pldi2010-schism-slides.pdf

For those who don't know, GC is often blamed- generally rightly so-
for the biggest reason why ruby is slow.

I have no affiliate with either of these links.

Andrew McElroy

> --
> Posted via http://www.ruby-forum.com/.
>

From: Abder-Rahman Ali on
Thanks Andrew. Helpful...
--
Posted via http://www.ruby-forum.com/.

From: Ammar Ali on
[Note: parts of this message were removed to make it a legal post.]

On Tue, Jul 13, 2010 at 7:42 PM, Abder-Rahman Ali <
abder.rahman.ali(a)gmail.com> wrote:

> Thanks Andrew. Helpful...
> --
> Posted via http://www.ruby-forum.com/.
>
>
Joe Damato did a good talk on this subject recently:

http://blog.envylabs.com/2010/07/garbage-collection-the-ruby-heap/

From: Rick DeNatale on
On Tue, Jul 13, 2010 at 11:40 AM, Bruno A <sardaukar.siet(a)gmail.com> wrote:
> On 13-07-2010 17:30, Abder-Rahman Ali wrote:
>> In the "Why's poignant guide to Ruby" book, it states the following:
>>
>> "If you can't get to an object through a variable, then Ruby will figure
>> you are done with it and will get rid of it. Periodically, Ruby sends
>> out its garbage collector to set these objects free."
>>
>> The point I'm not getting here is: "If you can't get to an object
>> through a variable..."
>>
>> Can you provide an examples that illustrates this?
>>
>> Thanks.
>
> a = [1,2,3].find {|x| x > 3}
>
> puts a
>
>
> In this (maybe too simple) example, x is garbage-collected once flow
> leaves the block.
>

Actually this is a bad example for several reasons.

1) variables don't get collected, object do.
http://talklikeaduck.denhaven2.com/2006/09/13/on-variables-values-and-objects

In this example x is a variable which over the course of execution
gets assigned to reference 1, then 2, then 3.

2) FixNums which are the only class of object assigned to x in this
example are never garbage collected, since they are immediate values.

Now what does GC really do? What it really does is make sure that any
object which can still be reached via a chain of references starting
with a root set of references WON"T be freed and reused before they
should be. This is really the primary job of a GC, the secondary job,
which most people think of what a GC does, is to allow the resources
used by objects which be provedn to be no longer be reachable from
that root set to be reused.

The root set comprised references such as:

Any constants in the outermost name space, (e.g. the object which is
referenced by say Array)
Any references currently on the activation stack of running threads.
These references will be things like method invocation parameter
values, local temporary variable values, and block argument values.

The reachable objects are those referenced by any reference in the
root set, any object referenced by any of the variables within those
objects etc. etc.

The effect of a variable going out of scope (e.g. a temporary on the
invocation frame of a method which has returned), is that that
reference is no longer part of the root set. but if there is at least
one other reference to the object in question reachable from somewhere
else in the root set then that object WILL NOT be reclaimed by a
bug-free garbage collector.


--
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale