I took a quick glance at egraphs-good (egg) and I can see the
advantage. I just wonder whether a setup like this could manage with
the complexity required for fully modelling J.
For now the calculus addon pretty much manually codes every possible
combination, each with its own, adhoc structure. I
I don't think we need to model all of J here. The derivative of ": is
not going to be a thing, for example.
Building up from a useful core is probably the way to go. And that
starts with rank 0 numeric functions (followed by their higher ranked
analogs). Having enough documentation to keep core co
I do not think there is any reason to use rust or ffi.
I think the concerns of a cas are different from those of a compiler, and no
attempt should be made to join the two. To give one pertinent example: the
type language used in a compiler should be total, whereas users should be able
to use
