Removing the relation ship between start / end state and beginning / end of processes ?

[jcelerier]

This was discussed at the Bergen workshop.

Pros:

• Simpler conceptually: the data is only at one place.
• Makes the code simpler: much less stuff to care about when changing a process, or a state.
• This would also fix some issues:
  • #469, #439
  • Some bugs where the content of the state is not correctly updated upon loading or copy paste & stuff like this

Cons:

• Less stuff is possible: for instance synchronizing the end of a process and the beginning of the next one. But maybe this would be better served by a "true" Sequence ?
• We have to rethink actions such as snapshotting, etc... and the workflow for working with automations.

[bltzr]

that's a very deep change, and I would recommend that we systematically try to weight all the consequences of this in all contexts where there is maybe that's something to figure out in parallel to specifying the "true" sequences, and implementing it along those (not before)

[jcelerier]

this would also allow an automation to finish on whatever value and not necessarily the value of its end state, if its parent is in a constraint with a trigger

[bltzr]

that surely makes sense, but we have to do it well, because if we just changed this without anticipating all side effects, this might break A LOT of things - and I mean, a lot

[bltzr]

I've been thinking about this, and I might have some ideas worth investigating in order to go the way you suggest - we'll also have a little brainstorm session with in July on this (and related) question(s)