when >1 root components are mounted over the same atom, update fails on all but 1 about om

[mccraigmccraig]

this may be naive, but is working for me on simple tests

https://www.refheap.com/22634

[swannodette]

We should probably do this by putting root functions into an atom around a set. When the render callback runs we should call all of the root functions in the set and empty it.

[mccraigmccraig]

i played with doing something similar, but then wasn't sure whether to scope the function-list atom by the state-atom, so that all root-functions for a given state-atom are ordered, or just make it global

[swannodette]

I don't think that render order matters in this case at all. Only atoms of roots that change will make into the render set.

[mccraigmccraig]

ah, i see : that's much better

[swannodette]

Fixed, 1b63c8f