Cellular Fluid

[rolyp]

Beginnings of a notebook-like notion of a program as a graph of cells. Ultimately cells might not be "static" (e.g. there might be conceptually one cell per point in the execution), but initially we just need the ability to build test cases (demos) that have multiple code fragments, each associated with an output view. Perhaps this would just consolidate the existing infrastructure in Demo.purs and turn that representation into a first-class language feature.

• 117, so we can load data directly in example code

• 521

• [ ] demo script specifies intermediate values to be visualised (e.g. by navigating trace?)

• 485, to visualise code "between" visualised values (by visualising trace, not program?)

[rolyp]

Some whiteboard notes from today, lots of our current todo's are somewhat related to this issue:

[rolyp]

I’m inclined to say this is an orthogonal requirement after all – linked inputs and linked outputs scenarios can easily exist within a “single” program (for linked outputs we just need some way of composing visualisations).

[rolyp]

Prezi‘s “nested presentations” could be a useful source of ideas.