On Thu, Jul 05, 2007 at 02:41:30PM -0700, Craig Latta wrote:
Hi JJ--
(Is it just me, or do all your messages come through without
newlines, even in the quoted material?)
Just as an observer from the side...
What's holding you back?
...isn't it the case that Pavel is trying the make the smallest possible code base *with current Squeak* while you are making lots of modifications to how the environment itself works?
That's one way to put it, I suppose. However, I suspect there isn't
an easy definition of what "current Squeak" is after you've done anything to it, unless your goal is to end up exactly where you started. Do we really want to end up where we started?
And if that's the case, would they still be cross purposes? I would see it more as "low hanging fruit" (so to speak) vs. "the whole shabang", no?
No, that's not how I see it. There's more involved in the value of
that fruit than the mere fact it hangs low. :) I think the amount of duplicated work, for results that aren't as useful, makes it something not worth doing that way (mostly because we are strapped for time and other resources). Having a short-term-gain mindset at all times will cause the total effort to be much harder and take much longer. I'm sorry if this sounds harsh (it sounds harsh to me, you don't need to convince me of that :). Despite that, I think it's still best to speak plainly here.
I don't see duplicated work going on between Pavel and Craig, only similar results.
Pavel, on the one hand, is finishing up the packaging effort that started in 3.9 (or earlier; I wasn't here before that). The goal is to split the monolithic image into packages that load and unload cleanly. 3.9 got almost all the way there; all that remained was Morphic. The goal is creating packages that work with the current tools (Monticello).
Craig, on the other hand, is building a system better able to handle packaging. Monticello was strapped onto Squeak at a high level, but is not usable throughout. Spoon, on the other hand, is the love child of Squeak and Monticello, with growth and modularity built right in, all the way down to VM support where necessary.
Thus Pavel is dividing up the image into much more manageable chunks, and Craig is providing a system for growth. Both efforts end up creating a small image without any extension modules, but the work to get there was very different, and was not duplicated effort.
As soon as spoon learns to walk, I daresay the first thing it will find to play with is the packages created using Monticello, including those made by Pavel.