On Wed, Aug 30, 2017 at 02:49:54PM -0700, tim Rowledge wrote:
> 
> > On 30-08-2017, at 12:14 PM, Eliot Miranda <eliot.miranda at gmail.com> wrote:
> > 
> > Hi Tim,
> > 
> >> On Aug 30, 2017, at 10:39 AM, tim Rowledge <tim at rowledge.org> wrote:
> >> 
> >> Whilst thinking about this let us please keep in mind leaving some hooks for use cases where there is no interrupt-type incoming events - older OSs (<cough>RISC OS, Mac OS 9, Windows pre-whatever) or bare-metal systems with essentially no OS. I don???t think either case should prevent us moving forward but at least leaving open, clean, avenues for them to keep working would be nice.
> > 
> > IMO, leaving it up to the VM to decide what to do when there's nothing to do allows better support on those other OSs and bare metal than leaving it up to the image to provide some catch all hack like the background process calling relinquishProcessorForMilliseconds.
> 
> No argument there. I won???t even argue that RISC OS is an important case these days, and that???s possibly the last in-use OS that requires the vm to be ???in charge??? whilst running and to voluntarily surrender control in order to keep things going. I???d just like to make sure there is a small flag waving somewhere reminding us that there are systems we might want to support that are not unix/windows/mac.
>

I for one think that RISC OS (or any other non-unix platform, take your
pick) remains an important use case. I personally think very highly of
Unix, and as the author of OSProcess I have taken an active role in making
platform specific functions available to Squeak images. But all the world
is not Unix. Never was, never will be, and never should be.

Smalltalk stands for a different way of thinking. It pains me every time
someone insists on adding a new platform-specific feature just because it
exists on Unix and maybe-sort-of works on Windows.

So bravo for RISC OS, SqueakNOS, SqueakJS, and Android VMs, keep them coming.

Dave