fix(gnovm): correct type for shift expression #1775
Conversation
…terface_comparison
ltzmaxwell
changed the title
ltzmaxwell
changed the title
|
Hi @thehowl , this one is good to review too. |
There was a problem hiding this comment.
Nice work. I think there are a few things that need fixing. I appreciate all of the tests. The Go spec has a some nice cases that maybe we could also include as part of the tests: https://arc.net/l/quote/offplfsv
Co-authored-by: Morgan <[email protected]>
There was a problem hiding this comment.
This PR greatly improves the behaviour of the shift operator. The tests are amazing and cover a lot of corner cases (several bugs also found in yaegi thanks to them) and match perfectly the results given by the compiler. Thank you for this effort. My remarks are minor and do no block approval. We should merge it ASAP now.
| @@ -0,0 +1,23 @@ | |||
| package main | |||
There was a problem hiding this comment.
This test is maybe legitimate but it has nothing to do with the shift operator. It should be renamed or suppressed.
|
|
||
| //s1 := []int{0} | ||
| //s1[0] = 1 + 1<<2 | ||
| //fmt.Printf("%T\n", s1[0]) |
There was a problem hiding this comment.
| //s1 := []int{0} | |
| //s1[0] = 1 + 1<<2 | |
| //fmt.Printf("%T\n", s1[0]) |
| @@ -0,0 +1,10 @@ | |||
| package main | |||
There was a problem hiding this comment.
This test has nothing to do with shift. Please rename it or suppress it.
UPDATE: this one is ready for review.
shift operator where first operand is an untyped bigint always results in a bigint is not resolved by #1426, it's fixed by this one.
=================================================================