Re: Reference Counting (was Re: Searching Method for Incremental Garbage Collection)

[hbaker@netcom.com (Henry G. Baker)]

In article <3bg6ci$18k@gamma.ois.com> beckwb@ois.com (R. William Beckwith) writes:
>An incr/decr only occurs when there is a change to the graph of
>objects.  If the graph does not change, then there is not need
>to incr/decr the counts.

This is fine, so long as every object is 'nailed down' directly or
indirectly to a root somewhere.  A problem occurs with newly allocated
objects that someone forgets to nail down, or forgets to deallocate it
after deciding _not_ to keep it.  A similar problem occurs when an
object is being destroyed, and it has been unhooked, but due to some
condition -- e.g., an error -- the program forgets to deallocate it.

An analogous situation happens in some GC systems -- e.g., Kyoto
Common Lisp.  If a compiled function returns a newly allocated object
which is then used as an argument to another compiled function which
then allocates storage, the GC can recycle the object, because it was
never 'nailed down' -- e.g., by storing a reference to it in the
run-time stack.  For this reason, one must used 'object' declarations
with some care in compiled code.

Henry Baker
Read (192.100.81.1) ftp.netcom.com:/pub/hb/hbaker/README for ftp-able papers.
WWW archive: ftp://ftp.netcom.com/pub/hb/hbaker/home.html
************* Note change of address ^^^        ^^^^