From: Will Pugh <willpugh@...>
Date: 2006-10-24T14:47:52+09:00
Subject: Re: Ruby's garbage collector...

Just in case you want to learn more, here is a pretty good survey of GC techniques.

ftp://ftp.cs.utexas.edu/pub/garbage/bigsurv.ps

    --Will

Will Pugh wrote:
> I do not know much about the Ruby Garbage collector, but I know Java 
> can easily collect circular references.  Most modern Java VMs use 
> Generational garbage collection.
>
> The only "garbage collection" technique I can think of right now that 
> fails to collect circular references is ref counting.
>
> I'd be interested in hearing about what kind of "pathalogical" 
> conditions you are talking about, but my guess is that they do not 
> actually prevent collection of inaccessible objects.  Normally, they 
> either increase response time or create additional memory overhead.
>
> The claim that "there is no perfect garbage collector" can be 
> generalized to "there is no perfect general allocation model".  Any 
> memory allocation contains a host of trade offs.  Java is actually 
> pretty advanced in a number of these, if you look at either the 
> JRockit VM or the more modern Sun VMs.
>
>    --Will
>
> Just Another Victim of the Ambient Morality wrote:
>>     Is there a name for Ruby's garbage collecting strategy?
>>     I'm on a web forum and the topic of garbage collection came up. 
>> There's a claim that you can never make a "perfect" garbage 
>> collector.  In other words, there will always exist (pathological) 
>> conditions where your collection strategy will fail to find 
>> inaccessible objects.  At least, that's my best understanding of the 
>> claim and it strikes me as... false.  I can understand that the task 
>> is non-trivial but simply impossible?  Can that be right?
>>     Furthermore, according to some, Java (a friendly Ruby competitor) 
>> is incapable of collecting, specifically, circular references.  I was 
>> just wondering if Ruby had the same limitation?
>>     Thank you...
>>
>>
>>
>>   
>