RE: Wiki Up

Hi Ralph,

Yes, I agree that neither VM is likely to supplant the other; it's a matter
of what's important to the developer.  Certainly, Strongtalk isn't going to
be running on all the platforms that Squeak runs on any time soon!

I haven't decided on what the right approach for a Squeak port would be yet.
The VMs and bytecodes are so different that I expect easiest thing to do
would simply be to try to find a cut point between the squeak VM and all the
code that doesn't depend upon the VM structure, and simply try to file in
and recompile all the non-VM code.

But of course, there is a huge body of problematic code, such as the byte
code compiler, image manipulation, debugger etc., and that is where the real
difficulties lie.  Dan thought that he could do this port by himself in a
month if he had the time, but either he mis-estimated that, or he isn't a
mere mortal like me (or maybe both!).

The bottom line is that a first cut at a port is likely to be fairly hacky,
and will almost certainly involve a temporary branch of the squeak code, and
probably the Strongtalk code too, but if we set a modest first target of
simply running reasonable squeak applications, rather than the debugger,
byte code compiler, etc., then hopefully we can get an excitement generated
to get some help figuring out what the real long-term plan should be.

Cheers,
Dave

> -----Original Message-----
> From: [email protected]
>
> Even if Squeak gets ported to the Strongtalk VM and is an order of
> magnitude faster, some Squeakers will prefer to use the existing VM.
> They find Squeak fast enough, and the current VM is easy to modify and
> to port.  Squeak runs on an amazing number of platforms.  However, a
> lot of people want a faster VM.  For example, I bet the Croquet folks
> would switch if Strongtalk were reliable and fast enough.  Most
> Squeakers don't hack the VM, and will be happy to get faster
> performance by just switching VMs.  Croquet only runs on a few
> platforms, so if Strongtalk ran on those same platforms it would be
> good enough for Croquet.
>
> Unfortunately, it will not be a matter of just switching VMs, people
> who want to use Strongtalk  VM will also have to switch images.
> Strongtalk will have a different bytecode format, a different set of
> primitives, and a different object format.  The usual way to make a
> major change to a VM is to use a program (whose name I forget) that
> traces all the objects in the old image and creates a new one.  To
> make a major change to bytecodes, you'd probably compute all the nes
> bytecodes in Smalltalk and store them somewhere (perhaps a dictionary
> from compiled methods to new bytecodes) and then when you write out a
> new image, you'd copy the new bytecodds instead of the old ones.
>
> In the long run, if Strongtalk is good enough, we could look into
> designing an image format that both VMs could use.  The Squeak VM is
> easy to change, so it might be easier to change it than to change
> Strongtalk, but undoubtedly compromizes would be needed.  In the short
> run, you should plan to have a program that converts a regular Squeak
> image into a Strongtalk-Squeak one.
>
> -Ralph Johnson
>
>
> 

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Strongtalk-general" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [email protected]
For more options, visit this group at http://groups.google.com/group/strongtalk-general?hl=en
-~----------~----~----~----~------~----~------~--~---