[PEAK] PyCells and peak.events

[Alexandre Garel]

Hello,
Thank you so much for this kind of mind-dump sharing. There are a lot of
interesting ideas in this post. Although I does not have your experience
in software design, I really feel in resonance with most of your Python
thoughts.

For me, locality of code (aka components) is a main condition to
maintainability. We really should provide mechanism so that each object
contains its own reactions to events (and not let outer objects have to
tell it what to do). Moreover all the wiring mechanism should be
automatic (like cells provide). For this to be usable it has to be
deterministic so test validate behavior, and as clear as possible for
programmers (not necessarily underlying mechanism but the effects they
provide). Incremental, features / aspect oriented (like rule dispatch
provide) is also necessary.
Concerning back propagation of errors, I guess the solution is
(nestable) transactions for actions sustained by a version mechanism for
each object (remember last version until commit, roll back when
necessary). Your "Constraints as Condition/Resolution pairs" post of
27/11/2004 is also really inspired.

I'd really like to see Peak becomes a community project with you as
benevolent dictator. I tell that but for myself, I have really little
time (strongly invested in important non computer related activities).

Keep the good work going,
Alex.