Forum: Ruby garbage-collection question OR ruby-gnome bug?

Announcement (2017-05-07): www.ruby-forum.com is now read-only since I unfortunately do not have the time to support and maintain the forum any more. Please see rubyonrails.org/community and ruby-lang.org/en/community for other Rails- und Ruby-related community platforms.
Cam (Guest)
on 2006-05-10 09:09
(Received via mailing list)
Hi,

My friend and I have been writing an application that makes extensive
use of gdkpixbuf's, but they never get de-allocated.  I'm not entirely
sure how the garbage-collector works internally in ruby, but my
program quickly grows to over 114M pretty routinely.  My friend wrote
a simple demonstration script:

require 'gtk2'
Gtk.init
path = ARGV.pop
raise "Pass a large image on the command line" unless path
puts "Look at how much memory I'm using, then press enter"
gets #1
buf = Gdk::Pixbuf.new( path )
puts "Now I've allocated the pixbuf, I should be using more memory."
puts "Hit enter to dereference the pixbuf."
gets #2
buf = nil
puts "Done.  Press enter to exit."
gets #3

Ok, here's the memory usage between various stages:

1:
1000     17271  0.3  2.4  15744  9312 pts/5    S+   23:03   0:00 ruby
pixbuf.rb

2:
1000     17271  0.3  4.1  22448 16228 pts/5    S+   23:03   0:00 ruby
pixbuf.rb

3:
1000     17271  0.3  4.1  22448 16228 pts/5    S+   23:03   0:00 ruby
pixbuf.rb

Now this example is much more trivial than the program we're trying to
right, but they're both having this issue w/ the pixbuf's never being
de-allocated.  Is this a known bug?  Is there something we can do to
force the GC to take care of them?

Thanks,
Cameron M.
Robert K. (Guest)
on 2006-05-10 11:08
(Received via mailing list)
2006/5/10, Cam <removed_email_address@domain.invalid>:
> path = ARGV.pop
>
>
> Now this example is much more trivial than the program we're trying to
> right, but they're both having this issue w/ the pixbuf's never being
> de-allocated.  Is this a known bug?  Is there something we can do to
> force the GC to take care of them?

I can only say something about your short demo piece because I don't
have the other code. It's quite unlikely that you see the GC kick in
before program termination here.  GC kicks in when the interpreter
decides to do it - not necessarily soon after an object has become
unreachable.  This is really only a problem if you have a long running
version and it keeps using more memory over time.  I suggest you
modify your script by removing lines with "gets" and put the whole
thing in a loop say 10,000 times.  Then watch how memory usage
develops.  If it's linearly increasing your chances are that there is
a bug in Gdk or you're using it the wrong way (i.e. do not release Gdk
resources as required by the lib).

HTH

robert
Cam (Guest)
on 2006-05-10 23:44
(Received via mailing list)
Hey,

On 5/10/06, Robert K. <removed_email_address@domain.invalid> wrote:
> resources as required by the lib).
I had thought about the GC not having time to free up the resources,
but I assumed that the GC would start doing its magic whenever  the
program was idle.  Anyway, the real program is much more complex than
this, and it should be having plenty of time for GC.  Is it common for
a library to require releasing of resources?  I will look into that
further, although I don't *think* i saw anything about that in the
docs.

Thanks,
Cameron M.
This topic is locked and can not be replied to.