Re: Best Ref-counting algorithms?

torbenm@pc-003.diku.dk (Torben =?iso-8859-1?Q?=C6gidius?= Mogensen)
Thu, 16 Jul 2009 09:39:51 +0200

          From comp.compilers

Related articles
[12 earlier articles]
Re: Best Ref-counting algorithms? torbenm@pc-003.diku.dk (2009-07-15)
Re: Best Ref-counting algorithms? cr88192@hotmail.com (BGB / cr88192) (2009-07-15)
Re: Best Ref-counting algorithms? cr88192@hotmail.com (BGB / cr88192) (2009-07-15)
Re: Best Ref-counting algorithms? cr88192@hotmail.com (BGB / cr88192) (2009-07-15)
Re: Best Ref-counting algorithms? cr88192@hotmail.com (BGB / cr88192) (2009-07-15)
Re: Best Ref-counting algorithms? gene.ressler@gmail.com (Gene) (2009-07-15)
Re: Best Ref-counting algorithms? torbenm@pc-003.diku.dk (2009-07-16)
Re: Best Ref-counting algorithms? bartc@freeuk.com (BartC) (2009-07-16)
Re: Best Ref-counting algorithms? gneuner2@comcast.net (George Neuner) (2009-07-16)
Re: Best Ref-counting algorithms? haberg_20080406@math.su.se (Hans Aberg) (2009-07-17)
Re: Best Ref-counting algorithms? haberg_20080406@math.su.se (Hans Aberg) (2009-07-17)
Re: Best Ref-counting algorithms? cppljevans@gmail.com (Larry) (2009-07-17)
Re: Best Ref-counting algorithms? lerno@dragonascendant.com (=?ISO-8859-1?Q?Christoffer_Lern=F6?=) (2009-07-17)
[15 later articles]
| List of all articles for this month |
From: torbenm@pc-003.diku.dk (Torben =?iso-8859-1?Q?=C6gidius?= Mogensen)
Newsgroups: comp.compilers
Date: Thu, 16 Jul 2009 09:39:51 +0200
Organization: Department of Computer Science, University of Copenhagen
References: 09-07-018 09-07-039
Keywords: GC
Posted-Date: 16 Jul 2009 18:33:52 EDT

George Neuner <gneuner2@comcast.net> writes:




> If you're dead set on reference counting, you should investigate 1-bit
> reference counting. I don't have URL's to give you but they should be
> easy to find. The programming language is designed so that most data
> are not shared and then you only need a simple flag to say whether the
> item is shared and so needs special handling.


You can get most of the benefit of 1-bit reference counts by using
linear-type inference. Basically, at compile time you find objects that
you can statically guarantee will have only one use _and_ (most of the
time) code the reuse directly in the compiled code instead of putting
the objects on a free list for later reuse. Non-linear objects are
GC'ed normally.


Torben



Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.