Hi Dan!
Hope Argentina was nice!
Dan Ingalls wrote:
Henrik Gedenryd picked up this work and carried it quite a bit farther into the system and sadly things got a bit too incompatible. It was a big frustration for me (and for him++) that we never managed to pull this off to completion.
While "almost" on the subject - did Henrik ever demo his "universal composition" work to you and his refactored implementation of BitBlt which IIRC turned into a very clean pipeline-ish thing with no performance penalty (since the runtime code turned more or less the same)?
We shared a hotel room at that OOPSLA and I recall he wanted to show it to you - but I am not sure he ever got the time. Since his "universal composition" stuff sounded quite intriguing at the time I was just curious to hear if you saw it and have any thoughts on it.
And yes, the Modules 3.3 branch in Squeak history is generally a sad chapter.
I hope, perhaps soon (now that we are again gathering speed to actually move Squeak forward into the future) - we can revisit those ideas and blend them with thoughts in Newspeak etc to see if we can eventually get something that "solves" the whole dependency/modules/packages/namespace arena.
It would be awesome if we could take at least *some* step in this direction.
Lesson learned though: the environment and tools MUST follow at the same time.
I also note that the same lesson has been re-learned regarding Traits...
regards, Göran
PS. Note to others - Henrik Gedenryd is unfortunately not with us anymore, he has passed away, unfortunately all too young.