From: Robert Klemme <shortcutter@...>
Date: 2007-04-23T22:25:07+09:00
Subject: Re: Tracking down a garbage collection problem

On 23.04.2007 11:03, Wincent Colaiuta wrote:
> On 21 abr, 16:50, Robert Klemme <shortcut...@googlemail.com> wrote:
>> I don't know what techniques exactly you employ but it seems indeed that
>> the cached data is held on to longer than intended.
>>
>>> As this is a largish, complicated project I don't even know where to
>>> begin to start investigating this. So really, I am looking for general
>>> information on techniques for measuring and exploring memory use and
>>> garbage collection in Ruby.
>> I am not sure, there might be an extension around.  Did you check with
>> the RAA?
>>
>> What I'd do here is a) code inspection b) create a finalizer with some
>> debugging output for your cache instance and see whether you see the
>> output.  If not, you know that your cache instance is still referred to.
> 
> Thanks, Robert. That is excellent advice. The method in question looks
> basically like this:
> 
> def parse
>   # do stuff
>   cache = MemoizingCache.new
>   # do more stuff
> end
> 
> I changed this to read as follows:
> 
> def parse
>   # do stuff
>   cache = MemoizingCache.new
>   $stderr.print "New cache id #{cache.object_id}\n"
>   ObjectSpace.define_finalizer(cache, lambda { |id| $stderr.print
> "Finalizer for id #{id}\n"})
>   # do more stuff
> end
> 
> And have confirmed that the cache finalizers are not being called. I
> haven't found out how the reference to the local "cache" variable is
> being kept alive even after it falls out of scope, but you've given me
> the tool necessary to at least find out what's happening.

Careful!  As Guy pointed out your code basically locks cache in memory 
itself.

You should rather do something like this:

def create_finalizer_for(id)
   lambda { puts "Final #{id}" }
end

and use that method to create the finalizer lambda.

def parse
   # do stuff
   cache = MemoizingCache.new
   $stderr.print "New cache id #{cache.object_id}\n"
   ObjectSpace.define_finalizer(cache, create_finalizer(cache.object_id))
   # do more stuff
end

HTH

	robert