Real Time GC
Sat, 7 Jun 1997 11:47:17 -0400
This sounds pretty good.
For the OOFS or persistant Lisp people, would you
think that implementing a non-persistant support
layer would be worth it?
> From: Bill House <email@example.com>
> To: LispOS Project <firstname.lastname@example.org>
> Subject: Re: Real Time GC
> Date: Saturday, 07 June, 1997 11:06
> ET wrote:
> > [snip decription of various GC issues]
> >Do modern real-time GC's suffer these same problems? I would think
> >the replicating GC would work nicely.
> What do you think of the L approach (see http://www.isr.com) -- being
> declare a separate heap that is not GC'd? Here's an excerpt from HB's
> regarding it:
> The L system utilizes a simple stop and copy garbage collector. The
> design strategy was too keep it simple and fast so that it runs
> unnoticeably in small applications with small heaps. The structure of
> L's implementation is such that multiple independent heaps can be
> maintained within a single address space, sharing all the code and
> data segments of Lisp proper. In this way super-critical portions of a
> system can be placed in heap where no consing is occurring, and hence
> there is no possibility that they will be blocked by garbage
> Bill House
> The views expressed are mine alone,
> unless you agree with me.