[gclist] object identity & copying collection

[Charles Fiterman]

At 10:59 AM 12/15/97 -0500, you wrote:
>>David Chase writes:

>It's also possible to generate tables telling the GC all about
>which offset pointers and other addressing-temporaries to repair
>when it moves a pointer, but I regard generation of those tables
>as an impediment as well; Eliot Moss and his minions have worked
>on this problem, and I recall that it was a bit of a pain.

I propose a simple compromise. Class descriptors should have
a single bit to distinguish leaf from non leaf objects. Leaf
objects can be allocated on separate pages and a conservative
collector doesn't even need to touch them in its mark phase.

Typical Java programs have lots of bitmaps etc which are
natural leaf objects.


			-  
Charles Fiterman		Geodesic Systems
414 North Orleans Suite 410	Phone 312 832 1221 x223
Chicago IL 60610-4418		FAX   312 832 1230
				http://www.geodesic.com

               ____
            .-" +' "-.    __,  ,___,
           /.'.'A_'*`.\  (--|__| _,,_ ,_
          |:.*'/\-\. ':|   _|  |(_||_)|_)\/
          |:.'.||"|.'*:|  (        |  | _/
           \:~^~^~^~^:/          __,  ,___,
            /`-....-'\          (--|__| _ |' _| _,   ,
           /          \           _|  |(_)||(_|(_|\//_)
           `-.,____,.-'          (               _/