[RFC] Add initial RFC for type casts. #1056
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1056 +/- ##
==========================================
+ Coverage 75.66% 75.84% +0.17%
==========================================
Files 439 438 -1
Lines 17771 17805 +34
==========================================
+ Hits 13447 13504 +57
+ Misses 4324 4301 -23
Continue to review full report at Codecov.
|
docs/rfc/004-integer-type-casts.md
Outdated
| (at least, not for production code). | ||
| Presumably, the behavior of type casts in those programming languages | ||
| is motivated by efficiency of execution, at least in part. | ||
| Since in Leo the input data is avaialable at compile time, |
There was a problem hiding this comment.
Typo -> "available". Also, do we want to distinguish between compile-to-IR-or-R1CS time and proof generation time in this discussion, or is that too tangential?
There was a problem hiding this comment.
I think it may be too tangential too at this level. (If you agree, you or I can click the 'Resolve conversation' button here.)
| let r_low16: u32 = r & 0xFFFF; // assuming we have bitwise ops and hex literals | ||
| let s: u16 = r_low16 as u16; // no value change here | ||
| ``` | ||
|
|
There was a problem hiding this comment.
The value-preserving casts alternative does not prevent the later addition of other operators that change the value and type, such as the re-interpretation between signed and unsigned. It might improve clarity of code to have a syntactic difference between value-preserving and value-changing casts.
Maybe we can start a discussion at this Wednesday's sync. |
Closes #1057.