[gclist] Garbage collection FAQ.
Charles Fiterman
cef@geode.geodesic.com
Tue, 27 Feb 1996 13:41:27 -0600
> > Should we have a garbage collection FAQ. It could also become part of
> > other FAQs on the net. What should the topics be?
> >
> > Algorithms.
> Basic:
I can write sections on the basic algorithms. Henry Baker thinks
reference counting can be revived will anyone take that question.
I have a weird mark-dont-sweep I'll put in because it illustrates
more advanced notions.
> Reference counting.
> Mark-and-sweep.
> Copying-compacting.
>
> Less basic:
> Conservative collection.
I can do this and have some really strange ideas I'd like to add.
We also have Hans Boehm in our group.
> Deferred reference counting.
> Generational.
> Baker's real time algorithm.
> Appel-Ellis-Li "real time" concurrent.
> Bartlett's conservative-compacting collector.
> Treadmill collector.
I've done this as C++ wrapper classes.
> Goldberg's tagless collection
>
> Assorted stuff:
> Blacklisting in a conservative collector.
> Hardware Support (Kelvin Nilsen's and others).
> One-bit reference counts.
>
> > OS Support.
> Zeroing out memory (I take it for granted, it need
> not be the case, especially on PCs).
> User-supplied fault handlers?
>
> Compiler support? (as opposed to "uncooperative environments" below)
> Lack of compiler support.
> Not hiding pointers.
> Nulling out dead pointer variables.
> Identifying pointers.
> Avoiding heap allocation.
>
> Folk myths?
> "Optimal" strategies that fail miserably.
>
> > Thread Support.
>
> > Garbage Collection for distributed objects.
I have an idea how this can work when some processors may not
be up.
>
> > Garbage Collection in uncoopertive environments.
>
> > Who will volunteer for various segments?
> > Who will run the show?
>
> I'd be interested in helping. Obviously, I've got strong opinions,
> but I've also been studying GC in "more conventional" (that is, less
> functional) languages for about ten years, on and off.
Take a few topics.
Who will take the executive job. I'm emotionally unsuited and
a poor writer. So I'm out.