... | ... | @@ -14,6 +14,19 @@ Here's the plan |
|
|
7. Write a loader that reads `link` code directly into the UM
|
|
|
8. Create a shell script that works as a "load-and-go assembler" by stitching together the assembler, `link`, and the UM.
|
|
|
|
|
|
# Ideas for when Norman's away
|
|
|
|
|
|
Possibly some basic functional-programming stuff for the front end:
|
|
|
|
|
|
1. Implement the generalized ANF from the MinCaml paper.
|
|
|
2. Implement inlining on the ANF.
|
|
|
3. Implement the ICFP'14 paper (from Francois Pottier?) on type inference and materializing a typed intermediate language.
|
|
|
4. Tiny specification, based on SLED (and using SLED syntax) for the
|
|
|
values of opcodes and their positions. Generate ML code for
|
|
|
building instructions and for poking at them. (Trying use the
|
|
|
[Toolkit](http://www.cs.tufts.edu/~nr/toolkit/) directly is a
|
|
|
dubious plan.)
|
|
|
|
|
|
# The next plan
|
|
|
|
|
|
Design of an instruction set for uScheme. |