WIP on Mutations Dec 2023

[lhstrh]

• revisit key chain logic
• static vs. dynamic scope
• set as method in sandbox, not the port itself
• set port for reactions (check that port is among effects)
• set port for mutations (add dependency if not present)
• unit tests for Sieve-like mutations discussed on whiteboard
• split reactor.ts into smaller files

[axmmisaka]

This has been paused forever, but just writing a memo for me here:
In one meeting Prof. Lee and others found that it might be a better idea to eliminate the ability for mutations to write to ports, i.e. the only effect a mutation can have is to carry out the 4 API calls outlined in the tech report.
This will not impact the ability of mutation to achieve its tasks because one can split an old mutation into a reaction and a mutation, one of which does the connection, and the other does the writing or nop.
In the code generator the syntax should therefore replicate what we have for reactions
mutation [<name>] (triggers) [<uses>] [-> <effects>] [{= ... body ...=}]
And unlike in reactions effects must be ports.