[Cuis-dev] Dynabook and education

[Douglas Brebner]

On 20/06/2020 10:41, Philip Bernhart via Cuis-dev wrote:
> Hilaire Fernandes via Cuis-dev <cuis-dev at lists.cuis.st> writes:
>
>> Oh yes this is wise. Even stronger, we do not have the choice given the
>> single thread nature of the VM, it will be a waste of resources to not
>> have several image running in parallel. I vaguely know that there are
>> inter process protocol for communication between process (IPC). Should
>> it be fast enough for communication between applications deployed in
>> different images?
> In this context I think objects should be as powerful or powerless as an
> Image. They both should be objects in the same regard. And when it comes
> to my preferences I think an actor based system would be an improvement,
> not putting the Images on different cores but putting the objects
> themselves on different cores.
I totally agree with this. I suggested turning Images into Actors first 
since they'd require much less invasive changes to the VM, probably much 
less annoyance to Eliot, and can be tested with normal image/vms running 
in parallel. Playing around with Little Smalltalk/SmallWorld among other 
things shows that it's possible to make tiny images from scratch that 
could be purpose made for specific tasks.
> And in the general thinking of mine I was referrering to the notion,
> that we humans tend to build a better now and not a better future.
> For example Alan Kay brought up the example of better horse carriage
> in the times of the automobile invention. It would not be the car which
> people wanted, but better horse carriages. Or in the time of Xerox not
> a Personal Computer, but just better Paper. Which is also, when you
> watch Alan Kay talks a frequent topic.

I have in mind a sea of objects/actors/images that can span multiple 
core and eventually, multiple machines a la the Internet. We could have 
actor/images of various sizes specialised for various tasks like a GUI.

I hope this makes sense.