[gclist] Name that hypothesis

Nick Barnes nickb@harlequin.co.uk
Tue, 03 Dec 1996 17:29:10 +0000

> At 12:40 PM 12/3/96 +0000, you wrote:
> >What is the name of the following hypothesis?
> >
> >	"Most references in a system point backwards in time, i.e. from
> >	 younger objects to older objects."
> >
> >In the GC community we all know this, and many of us use it
> >frequently, but we don't seem to have agreed on a name for it.
> How about "We remember the past not the future hypothesis."
> This is part of something more general. Programmers tend to write
> functions that return objects. Once these objects are created they
> are rarely modified. So a constructor can put references to older
> objects in a new object but if you never modify the new object only
> pass it around as a value its references can only point backward.
> How about the "Functional programming is the norm, side effects the 
> exception hypothesis."

Right. Clearly the Ordering Assumption (thanks to Dave Mason) is
related to this "Rare Modification Hypothesis": old objects are rarely
modified. The RMH is very important to some implementations of write
barriers, and implies the OA. But the implication doesn't go the other
way: there could be a few old objects, updated very frequently.

(maybe the RMH should be "few old objects are ever modified").

Nick B