miloignis/kraken
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Finally fixed let_test4. Let_test5 will require more thinking to see if it's even possible, and what macro like means etc

Browse Source
This commit is contained in:
Nathan Braswell
2021-10-18 00:46:39 -04:00
parent 923c4565fb
commit 92ac879c6d
2 changed files with 34 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
partial_eval_test.kp
View File

@@ -61,26 +61,6 @@
let3_test (read-string "((wrap (vau (let1) (let1 a 12 (wrap (vau (x) (+ x a 1)))))) (vau de (s v b) (eval (array (array vau (array s) b) (eval v de)) de)))")
let4_test (read-string "((wrap (vau (let1) (let1 a 12 (wrap (vau (x) (let1 y (+ a 1) (+ y x a))))))) (vau de (s v b) (eval (array (array vau (array s) b) (eval v de)) de)))")
; Ok, the post-refactor sticking point is
;
; after first eval of param ( marked_array true ( ( marked_array true ( ( prim_comb combiner(wrap_level: 1) builtin_combiner_vau(wrap_level: 0) ) ( marked_array true ( ( marked_symbol true y ) ) ) ( marked_array true ( ( marked_symbol true + ) ( marked_symbol true y ) ( marked_symbol true x ) ( marked_symbol true a ) ) ) ) ) ( marked_array false ( ( prim_comb combiner(wrap_level: 1) builtin_combiner_+(wrap_level: 1) ) ( marked_symbol false x ) ( val 12 ) ( val 1 ) ) ) ) )
;
; tries to finish the eval by unvaling & then partial evaling:
; [ [ vau [ 'y ] [ '+ 'y 'x 'a ] ] ( + x 12 1 ) ]
;
; This fails as it can't unval (+ x 12 1). Note the vau's not wrapped, so it won't actually partial eval after that, but it still dies first...
; This is where that is_val as an int might make sense...
; theoretically when the vau uses y and then strips it can sub in the stuff exactly, as subbing in itself counts as an evaluation.
; In general, stripping counts as a +1 to the is_val counter and we need to add evals or (array ...)/quote to get it to 0. In this case, it would work perfectly.
; The REALLY tricky part is that by allow it to go negative we have to remember what environment it needs to be evaluated in and make sure it's either the same environment
; or a sub environment that doesn't shadow anything...
;
; ALTERNATIVE: allow partial evals on things that contain negatives, but don't actually do the call, but allow the partial eval to go into the other
; parts, namely into the body of the vau above
;
; Also, it seems to be bailing even harder than it otherwise should be, as that above partial eval of let1, as limited as it is, doesn't show up in the final output
; This is due to the later? and closes_over_var_from_this_env_marked check in function call
;
;!!!!!!!!!!!!!!!!!!!!!!!!!!
; Which means we need TODO
;!!!!!!!!!!!!!!!!!!!!!!!!!!