Closed rokopt closed 1 year ago
(The latest CI failures are due to less-than
not being hooked in to end-to-end tests yet.)
I've added explicit type information to lamda:left
and lambda:right
for #53 .
Unfortunately, the type-checking is commented out at the moment, because it fails as follows:
Types should match for left s-1 s-1 BOOL
Those types should match, so the object-equality test must still be insufficient, at least for BOOL
. Maybe #71 would fix this?
I've added explicit type information to
lamda:left
andlambda:right
for #53 .Unfortunately, the type-checking is commented out at the moment, because it fails as follows:
Types should match for left s-1 s-1 BOOL
Those types should match, so the object-equality test must still be insufficient, at least for
BOOL
. Maybe #71 would fix this?
probably not, for now you can check past aliases, with the meta data patch being in, we can deprecate any pattern where we match on aliases, and instead have them be implicit in the code we call
All of the commits from this PR have now been merged to main
in other PRs, so I'm closing this one.
I'm making this pull request a draft because we want to either break it up into several new PRs or integrate bits of it with existing PRs or something. But I am creating the PR because I can't figure out how to test the code without CI;
asdf
doesn't seem to run newly-written tests no matter what I do (including deleting a bunch of files), at least for some unpredictable period of time; it eventually happens for reasons that I don't understand any more than I understand the reasons that it doesn't happen right away.The contents of the PR:
lambda
from implicit to explicit, to make comparing the Lisp code to it easierso0
andso1
cases ofcurry
that were using the wrong argumentsabsurd
STLC term (see comment https://github.com/anoma/geb/issues/53#issuecomment-1421093312 to #53)left
andright
STLC terms for #53