Hi Tony--
> I'd love to follow your ongoing work on Spoon more closely. How are
> you structuring it?
It's structured as in the latest release[1], as a master image with
development tools and a slave image which is minimal.
> Could your intermediate "snapshot" work-in-progress builds be made
> available to the public somehow?
It's difficult because of the very problem I'm trying to solve
(modularity). My master image has several other of my projects in it,
not all of which are appropriate for public consumption. Creating an
appropriate snapshot is currently time-consuming. Once I have
bootstrapped Spoon's module system, I plan to keep all snapshots
publicly available all the time.
As I mentioned in my most recent progress report, I plan to make a
release soon, despite the burden.
thanks,
-C
[1] http://ftp.squeak.org/Spoon/spoon1a12.zip
--
Craig Latta
http://netjam.org/resume
Hi--
I have successfully removed class SystemDictionary (and its sole
instance, "Smalltalk") from the minimal object memory. There is now no
global state per se, only a few objects that the VM knows about via the
"special objects array". As I described before[1], each method literal
for referring to a class is the responsibility of the class itself
(stored in the "name" slot provided by Class).
Now I'm moving on with creating a "hello world" module (one that
just draws anything to the display). I plan to make another release when
that works.
more to come,
-C
[1]
http://lists.squeakfoundation.org/pipermail/spoon/2006-July/000124.html
--
Craig Latta
http://netjam.org/resume
Hi PhiHo--
> Tim wrote:
>
> > > Maybe it would be better to just abandon the current image lineage
> > > and jump ship to build this purely on top of Spoon?
>
> and you responded:
>
> > Yeah, let's do that one. :)
>
> I know you are seriously smiling here.
>
> Yeah, let's do that one !!!
>
> I don't know if you are serious enough to use your Spoon to scoop out
> from the release Squeak 3.9 the KernelImage pioneered by Pavel and
> post the system and instructions for reproducing it...
Yes, I'm completely serious; but, unlike what I'm making, I don't
think Pavel's image is really a kernel (it's not minimal). I think it's
great if his work shows a way to delineate particular subsystems, though.
In general, I think "stripping" is a losing strategy. There's only
one strip that matters for making a basis artifact, and that's the one
that gets to an absolutely minimal core. I hope no one has to do it
again after I'm done. From then on, the system should be composed of
modules which can just be told to unload. The modules should worry about
handling the ramifications of their dependencies (asking for human
intervention only when absolutely necessary). It should be possible to
compose any desired system by loading modules into the minimal core.
As for the 3.9 image, that's just one of several sources of
behavior that I plan to imprint onto the Spoon object memory. I plan to
release Naiad for each one, so that people can imprint from them.
-C
p.s.
I'm on the verge of another milestone: I have successfully removed
all references to SystemDictionary and its sole instance from the
minimal object memory, and my remote browsing tools work without
referring to them as well. I'm about to remove them.
--
Craig Latta
http://netjam.org/resume
