[Squeakfoundation]The Natives are Restless
Fri, 25 May 2001 01:36:42 -0400
Paul Fernhout wrote:
(much discussion of the "vision" thing, modularity, complexity management, VisualWorks-lite, etc., snipped...)
> In other words, and obviously it's not quite this bad, but, if the
> complex house of cards is near to going crashing down in flames without
> some major changes, why should it be the Squeak Foundation that takes
> the heat? Perhaps instead a "burn the disk packs" approach might make
> more sense (such as Jecel Assumpcao Jr is doing with a new VM approach).
> That way a new system with a clear title for every contribution,
> organized in a modular way, with additional complexity management tools,
> could be built from the ground up in a distributed fashion by the Squeak
> community. Obviously, though "burn the disk packs" doesn't build that
> much on the existing Squeak community or knowledgebase, and so who would
> the developers or users be? Still, let's assume complexity management
> is, say, 1980 technology. How do we get it, and how should this fit into
> the SqF agenda?
A "burn the disk packs" approach for the Squeak Foundation sounds a bit impractical to me, although not totally without merit. At least, it would be a very large amount of work, which would imply that a significant number of people would need to get behind the idea, and I haven't heard anyone else mention this as a possible vision for the SqF.
On the other hand, the Stable Squeak effort is a partial move in this direction already. It addresses modularity, it will have some additional complexity management tools (for managing the modules, at least), but it doesn't address the "clear title for every contribution" issue. Given the effort behind Stable Squeak, I'm guessing it's probably unrealistic to hope for another even larger start-from-scratch effort, but I certainly wouldn't stop anyone. :)
I'm wondering if requiring a clear title for every contribution might be an overly paranoid/conservative approach, but I'm not a lawyer and don't claim to know what's best in this situation. (I wonder if it might be sufficient to review just the significant contributions from third parties, and not stuff like bug fixes?) However, given that Stable Squeak will resolve the modularity issue, in theory it might be possible to tackle this with each module, starting with the kernel module (which I would guess has a smaller proportion of third party contributions than other modules).
Regarding complexity management tools, I agree that more needs to be done... changesets should not be the ultimate means of distributing code/applications. But I think this can evolve without a "burn the disk packs" approach.
You mentioned the least-ambitious option for SqF, supporting something akin to a VisualWorks-lite... I guess I don't see it ever being quite that limited, since Squeak Central (and other Squeakers, researchers, etc.) will always be out there doing blue-plane work, at least some of which would make it back into the more stable/commercially-oriented version, whatever that may be.
Anyway, enough rambling from me, but I agree that a bit of discussion on the vision thing is worthwhile. (although perhaps it's a bit premature to discuss the Stable Squeak side of things in depth, since it's not yet released)
- Doug Way