casting integer literal to float is unnecessary #3877
Conversation
flip1995
added
the
S-waiting-on-author
tests/ui/cast_size.stderr
Outdated
|
|
||
| error: casting usize to f64 causes a loss of precision on targets with 64-bit wide pointers (usize is 64 bits wide, but f64's mantissa is only 52 bits wide) | ||
| error: casting integer literal to f64 is unnecessary |
There was a problem hiding this comment.
This warning shouldn't change, as for example 9999999999999999_usize != 9999999999999999_f64
There was a problem hiding this comment.
Yes, it is.
But 9999999999999999_usize as f64 == 9999999999999999_f64, so the suggestion equal to orgin code.
Maybe we should check mantissa bits wide for integer literal case.
Then add the test case mentioned above.
Is it ok?
There was a problem hiding this comment.
Maybe we should rewrite some test case.
The error type doesn't match name of test case.
There was a problem hiding this comment.
I fixed some test case to keep original error type.
There was a problem hiding this comment.
9999999999999999_usize as f64 == 9999999999999999_f64
That is correct, but the previous warning explicitly warns you about this kind of cast, since you lose precision with it. The new suggestion would indicate to the programmer, that this cast is always ok and he can change the code.
|
For the purpose of easier reviewing: Could you add new commits for a change and avoid force pushing? You can then squash the commits later |
|
I think this looks good now, but I have to take a closer look at it this evening. |
|
Thanks! @bors r+ |
|
📌 Commit d9dd008 has been approved by |
casting integer literal to float is unnecessary fix issue #3842
|
☀️ Test successful - checks-travis, status-appveyor |
fix issue #3842