Appears to work just fine … for those who cannot run this in their system, here it is in SqueakJS:


On Sun, Jul 26, 2020 at 12:02 David T. Lewis <> wrote:
Hi Eliot,

On Sat, Jul 25, 2020 at 08:33:38PM -0700, Eliot Miranda wrote:
> Hi David,
> On Sat, Jul 25, 2020 at 10:04 AM David T. Lewis <> wrote:
> > A couple of years ago, I put a copy of a Squeak 5.2 image in V3 object
> > memory format on the server, and I linked to it from the
> > project page at That link
> > has gone dead, and I think I recall someone noticing it a recently.
> >
> > I still have the files, should I put them somewhere on
> > I was thinking of putting them under the "various_images" folder:
> >
> >
> >
> > I also have a V3 image that is up to date with the Squeak 5.3 release,
> > so I could put that on line as well.
> >
> > These images will run under an up-to-date interpreter VM or a Cog VM. They
> > may be of interest to people doing VM work, benchmarking, or as a reference
> > to understand the image-side differences between Spur and V3.
> >
> This is definitely useful.  My only concern is if this moves trunk towards
> having code that supports both v3 and Spur.  I don't want the complexity or
> the unnecessary overhead that entails.  So can we please keep a series of
> patches to keep 5.x/6.x v3 going and not change trunk to accomodate?

Not to worry, there are no changes to trunk. The project page for explains what
I have been doing.

In general, I have been able to keep MCM file numbers in sync between
trunk and the packages in TrunkUpdateStreamV3. So for example if you
have a Squeak 5.2 image with Compiler-eem.394 then you can look at
Compiler.V3-dtl.393 in the TrunkUpdateStreamV3 repository to see the
differences. Please don't expect this to be perfect, it definitely
is not.

I should note that Juan has been maintaining Cuis for Spur64, Spur32,
and V3 from a common code base. So if someone is interested in how
much complexity is added by supporting this, then Cuis is the best