On Feb 7, 2004, at 6:09 PM, Tobias Nurmiranta wrote:
> To convince Tim, or others, that its a nice algorithm, is a later 
> task. :)
I wish you luck. If you're doing this at a college there are much more 
important people than me to convince...

> To make it really efficient in practice, it probably will need 
> integration
> with the OS's paging mechanism.
.. and there is what I suspect is the ugliest bit. Typical pagers for 
OSs seem to obsess on memory pages of a fixed size. Given how most of 
the hardware I've ever bothered to read about does things this is not 
at all surprising.  Oddly enough one of the CPUs that doesn't obsess 
over fixed size memory pages is the ARM; at least a few versions have 
an MMU that can have wildly variable page sizes. I think Apple 
initially came up with the implementation back in the days of the 
Newton.

If you haven't already done so, you should look at the Mushroom 
archives; searching for Mario Wolczko, mushroom and smalltalk ought to 
find something.

tim