On Jan 31, 2008 1:27 PM, Andreas Raab <andreas.raab at gmx.de> wrote:

> Laurence Rozier wrote:
> > This is true and you make many other good observations. Since these tend
> > to apply within just the world of Squeak and there exists a spectrum of
> > funded entities it would seem that the prospects for improving the
> > situation are better there. In order for that to happen some small group
> > of people will have to decide that it's in their collective interests to
> > establish and maintain a Squeak kernel. The only way for the Squeak
> > Foundation to do this is to convince at least 3-4 of the highly visible
> > projects - say Squeakland, Croquet, Seaside to commit to a common
> > foundation.
>
> But does *anyone* even have the slightest idea what that would entail?

Yes, I do but the price(both human and financial capital) is probably higher
than any one entity can justify and it's not likely that the pain can be
distributed evenly. Given a sufficient pile of euros, I know that this can
be done. It's not a computer science or software engineering project but
rather an engineering systems problem.

>
> It sounds great as a theory but in practice I have never seen a setup
> that has worked across significantly different code bases. The only
> working granularity in my experience is the image and in practice that
> means that unless these projects share a common image (which I find
> highly unlikely given that it would imply decisions about, for example,
> the scope of Morphic supported in it) it seems almost impossible to get
> something like what you are describing going.

It is possible to craft an evolutionary plan to get to a viable common
image, one that meets the needs of different projects. Similarly wrt VM
issues. Initially some may have to do a bit more work than others and
backward compatibility will have to be sacrificed in more cases than people
would like but the premise is that the ends justify the means. Here's a
thumbnail sketch of my scenario for addressing this:
-Identify a minimal image from which each project could be built
-Identify the major conflict areas
-Identify what improvements to current tools(Spoon, Tweak, VMMaker, Delta
Streams, Monticello and whatever else might help) are needed to build each
project from the minimal image
-Determine the implementation goals and phases(2-3 max)
-Update the tools
-Use the tools

Clearly there are lots of difficulties involved with this but I don't see
how any of the problems won't yield to a determined pursuit along these
lines. Plus such an effort would get Spoon or some derivative of it baked in
which IMO is one of the best things that could happen to Squeak. There are
many important areas that need to be explored. Your Morphic example is one -
maybe it would be more effective to migrate from Morphic to Tweak and
manually transition older work work - or leave it behind.

In another life, I was faced with a gear housing for a precision spacecraft
mechanism whose main shaft bore was too large. Conventional wisdom says you
can't make holes smaller and while good engineers know that in some
situations you can build up material on the surface of this wasn't one of
those times. There was no way to get new ones made and meet the launch
schedule so I came up with a way to make the bearings that went in the bore
larger. We made a precision tool that held and cooled the bearing while
heating a thin precision-machined ring and slipping it over the bearing.
Making and ring and tool was not a trivial project but it was doable and the
stakes were high. I think the stakes are very high for the three
projects(perhaps adding Sophie) I mentioned. While they're all likely to
survive, if they don't expand their niches significantly, survival is all it
will be and having experienced the difference over the past two decades I am
in favor of paying the price for expansion. While Second Life isn't going
anywhere<http://croquet.funkencode.com/2008/01/31/why-second-life-will-get-a-third-life/>,
there's still a significant window of opportunity for greatly expanding the
presence of Squeak via Croquet.  With well-integrated e-Toys, Seaside and
other Squeak goodies Croquet will be strengthened while providing an
expanded market for other Squeak projects. Other Smalltalks would have more
incentive to join the party too.

Cheers,

Laurence

>
>
> Cheers,
>   - Andreas
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20080131/35965b0d/attachment.htm