src/nockjs/TODO · bridge

on c19b2a8 help sign in

42 lines 1.7 kB history raw

1* a noun is an atom or a cell. DONE

2* mugs. DONE

3* unifying equals. DONE

4* dorky embedded DSL for nouns. DONE

5* some kind of real hashmap (needed for memoization at least). DONE

6* nock->js compiler (THE REALLY INTERESTING PART). DONE (at least, v1)

7* decrement. DONE

8* port random-noun tests to tape-check. DONE

9* jam/cue for pill reading (makes testing easier). DONE

10* add. DONE

11* jets (basic binding). DONE

12* ackerman. DONE

13* atom-gates.pill

14* ALL jets (ref: jaque)

15* boot ivory pill

16^--- call urbit library functions from js! great success!

17* boot fakezod (brass pill) to dojo

18* prevalence (on-disk in node, IndexedDB in brower)

19* expose basic vere-slave interface as a library

20 i.e. send me pokes, i handle all persistence and send you io

21^--- candidate vere backend

22 urbit in the browser

23* urbit over webrtc?

25At this point, we have a working vere backend, as well as a full

26"urbit in the browser". Performance is unlikely to be as good as jaque,

27but will likely be quite accepatable (especially for in-browser things).

29At the very least, using an ivory pill to compile and run hoon from the

30browser will ease client-side programming tasks.

32--- future implementation ideas

34it would be feasible to add profiling code to the compiled javascript,

35specifically in callsites (9), so that the 9 would run say 100 times and

36then recompile itself with a check and a direct call to the most called

37battery. this would also enable loop finding in tail position where the

38most called battery is the one being compiled. (important here not to

39use the 2 compiler cache, but to have a generated call site per battery)

41it would be cool to profile that single optimization, all other things

42being equal. it should be the most important one.