[ENH][Modules] Should we port Ginsu to 3.1 so that people look at it?

Dan Ingalls Dan at SqueakLand.org
Fri Sep 14 23:07:16 UTC 2001


Stephane -

>I have the impression that really few people look at the work of Joseph (I
>maight be wrong)

This is probably the case.  For any given deep topic on this list, usually only a small core of people get seriously into it (VMMaker, Linda, 3D, JPEG, ... you name it).

I looked at it, filed it in and measured it.  I think about a half dozen other people (plus Paul and G–ran) have gone this far.  I hope as many are also taking a look at what Henrik has done.

>I have a proposition: should we port into Squeak3.1 what joseph made so that
>people can look at it?

Well I feel that it's late for this proposition.  I would have to dredge through all the old mail, but I think we talked about that way back when the work was going on in the Squeak World Tour.  Then a month ago, when the discussion got going again, we asked (I think several times) for what it would take to bring it into 3.1, and who would do it, as well as more important things such as was it separable into a simple kernel, how small could it be, and how would the normal process of programming in Squeak be affected.  Space overhead is a significant issue.  I can get into this more specifically in another message or you can investigate it on your own.

I have not felt a positive force in the direction of answering those questions, nor getting it ported, other than Paul and G–ran trying to be good citizens.

We're a month down the road now, during which time Henrik has taken it upon himself to describe an approach and to actually get something small and simple working.  He has made it work in 3.1, and has already made several of the tools work with it.  It looks clean and general, and it looks like he has paid attention to suggestions from both Allen and Les, as well as what made sense in the kernel of Joseph's work.  This is a positive community process.  It inspires my enthusiasm.

I think Joseph has pulled off a major piece of work, and I expect that a number of people will want to use his system for its noble strengths.  However I think we would need to feel more collaboration and see the possibility of a much smaller kernel to incorporate it all in the Squeak base.  Conversely, I think most of the really hard work that Joseph has done could fairly simply live on top of a module system such as Henrik's, and we would have the best of both worlds.

	- Dan
-- 




More information about the Squeak-dev mailing list