Rather surprisingly, all the treeshake tooling is done in just about 2
hours of work. From now on it'll be easier to update stats.txt. Observations:
a) Binaries are tiny compared to conventional stacks. Tens of KB.
b) ~80% of binaries are tests and unused libraries in all my apps.
c) ~75% of LoC in SubX sources are tests or comments.
Start of a new script called treeshake to emit stats for minimal line counts
and binary sizes for all apps.
It doesn't actually do any dead-code deletion yet. But it does build and
run all apps successfully. (Except apps/mu; we'll ignore that for now.
It's probably not being disciplined about identifying internal labels.)
A couple more primitives now working. In the process I ran into an issue
with some buffer filling up when running ntranslate. Isolating it to survey.subx
was straightforward, but --trace ran out of RAM, and --trace --dump ran
out of (7GB of) disk. In the end what helped was just repeatedly inserting
exits at different points, and I realized there was a magic number that
hadn't been turned into a named constant.
Binaries are now identical again.
There's a little hack here that we should clean up at some point. But it
requires more thought.
Ordering compiler phases is hard. So far we're only at the start of the
slippery slope into that abyss.
Support binary operations with reg/mem and reg operands.
Everything is passing. However, the self-hosting translator now generates
some discrepancies compared to the C++ translator :(