[9fans] limbo's GC

[9fans] limbo's GC

[David Tolpin]

I have read the article and I am convinced. It is an algorithm different from
that I had in mind, but is efficient enough to support general-purpose garbage
collection.

I have been misled by limbo's papers and manual. 

Can I rephrase it as that I can use arbitrary data structures exactly as I would
with lisp or ML or any other language with automatic memory management. And
that I can rely on instant deallocation of resources if the data structures
behind resources are acyclic?

David

Re: [9fans] limbo's GC

[Charles Forsyth]

[-- Attachment #1: Type: text/plain, Size: 322 bytes --]

yes.

>>that I can rely on instant deallocation of resources if the data structures
>>behind resources are acyclic?

yes.  actually, that's one slight difference with Java (as i knew it, anyhow).
a Java implementation might deallocate acyclic things immediately
but the language doesn't guarantee it.  Limbo does.

[-- Attachment #2: Type: message/rfc822, Size: 2410 bytes --]

From: David Tolpin <dvd@davidashen.net>
To: 9fans@cse.psu.edu
Subject: [9fans] limbo's GC
Date: Tue, 2 Mar 2004 14:08:40 +0400 (AMT)
Message-ID: <200403021008.i22A8eJv072216@adat.davidashen.net>


I have read the article and I am convinced. It is an algorithm different from
that I had in mind, but is efficient enough to support general-purpose garbage
collection.

I have been misled by limbo's papers and manual. 

Can I rephrase it as that I can use arbitrary data structures exactly as I would
with lisp or ML or any other language with automatic memory management. And
that I can rely on instant deallocation of resources if the data structures
behind resources are acyclic?

David