Upgrade exponentiation to unified operator #7153
Conversation
| let int32_pow ?comment (e1 : t) (e2 : t) : J.expression = | ||
| match (e1.expression_desc, e2.expression_desc) with | ||
| | Number (Int {i = i1}), Number (Int {i = i2}) -> | ||
| int ?comment (Ext_int.int32_pow i1 i2) |
There was a problem hiding this comment.
Have you checked that this is semantically equivalent to js power, or whatever Pow does below?
There was a problem hiding this comment.
This is different from js pow, especially when overflow. If I leave js pow out, wouldn't it cause undefined behavior in ReScript, like values greater than max_int?
There was a problem hiding this comment.
Hmm, seems it is already possible by n ** n
There was a problem hiding this comment.
The intended question is: does the case with constants behave exactly like when 2 variables with the same constant values are passed?
There was a problem hiding this comment.
added more tests to verify it
cometkim
force-pushed
the
exp-operator
branch
from
6253a22
to
8af675c
Compare
Also make its output to use ES7 exponentiation (`**`) operator. `**` is more concise, faster than `Math.pow()`, works well with all numeric types include bigint. We were already using it for bigint, now for int and float too.
cometkim
force-pushed
the
exp-operator
branch
from
2ac5b3a
to
73dedfb
Compare
|
I think The problem is that to match the semantics we need to perform the modulo operation ourselves, which is quite expensive at runtime. |
|
Well, I guess I'll have to propose new int semantics to fix it, but that would be a pretty breaking change. It seems very weird now that all other operations use int32 coercion, but only multiplication has a different behavior. |
What is the difference, in short? |
|
Oh you mean |
|
(discussing in rescript-lang/rfcs#1) |
Also make its output to use ES7 exponentiation (
**) operator.**is more concise, faster thanMath.pow(),works well with all numeric types include bigint.
We were already using it for bigint, now for int and float too.