[gclist] Re: gclist-digest V1 #2

[Richard Jones]

>From: Amit J Patel <amitp@Theory.Stanford.EDU>
>Date: Fri, 8 Mar 1996 12:50:25 -0800
>Subject: Re: [gclist] Destructor FAQ
>
>Would it be feasible for the garbage collector to maintain a list of
>functions that must be called AFTER a collection?  That would the
>right time to check for cleanup, and it would also free me from having
>to explicitly call the cleanup function.  Every time new data is
>added, it could register a cleanup function with the GC.  (Is this
>getting back to finalizers?)
>

This can certainly be done. In fact the SRC Modula-3 collector provides
hooks to register routines that must be performed either before or after
a garbage collection.


>------------------------------
>
>From: boehm.PARC@xerox.com
>Date: Fri, 8 Mar 1996 13:44:34 PST
>Subject: [gclist] Deferred reference counting
>
>David's FAQ draft currently defines deferred reference counting as the
>technique of deallocating unreferenced structures lazily.  Is there a
>consesnsus that this is the one and only definition?
>
>If so, is there a consensus as to what we should call the Deutsch-Bobrow
>technique of not including the stack in the reference count?
>

I agree with Hans. I would suggest that the term "deferred reference counting"
is best restricted to techniques like Deutsch-Bobrow. The former technique
(first suggested by Weizenbaum in 1963) might be better termed "lazy freeing"
or "lazy deallocation".



  

Richard Jones

===============================================================================

Computing Laboratory                           Room SW107
University of Kent at Canterbury               Telephone: 01227 764000 ext.7943
Canterbury,                                               01227 827943 (direct)
CT2 7NF, U.K.                                  FAX: 01227 762811

===============================================================================