Better support for -0.0 #4228
Better support for -0.0 #4228
Conversation
VincentLanglet
force-pushed
the
poc/12225
branch
from
7fe1a6a to
9ae0da1
Compare
VincentLanglet
force-pushed
the
poc/12225
branch
from
9ae0da1 to
f6132e5
Compare
VincentLanglet
force-pushed
the
poc/12225
branch
4 times, most recently
from
d183265 to
a683afb
Compare
|
This pull request has been marked as ready for review. |
|
in phpstan/phpstan#12225 (comment) you mentioned a always-true check. will this error be reported after this PR and needs to be test-covered? |
No, but it's finally another issue because you already has the issue with something like |
VincentLanglet
force-pushed
the
poc/12225
branch
from
a683afb to
369a0f3
Compare
|
What if instead of all of these changes, we'd just change |
|
The other solution I had in mind for the issue indeed was to "just" update ConstantFloat::toString but it has the drawback that would have been considered as While with this implementation I have the behavior Your call. |
|
Yeah, I'm probably fine with that. We could improve that in the future by adding |
|
Closing in favor of #4278 then |
4 participants
This could be a solution for the issue phpstan/phpstan#12225
0.0and-0.0since they have a different toString value=== 0.0like=== 0.0 || === -0.0== 0The other solution I had in mind for the issue was to "just" update ConstantFloat::toString but it has the drawback that
would have been considered as
'0.0'|'-0.0'while here the constant was really known.Closes phpstan/phpstan#12225
What could be done to have less changes, would be to modify UnionType::describe to merge 0.0 and -0.0.
But I feel like it's better to be explicit, (since 0.0 and -0.0 have a different behavior) and also, there is already some
-0.0displayed by PHPStan like in the following snippet (and you'll see the already existing inconsistency).https://phpstan.org/r/0b3cdd4f-b228-4272-a685-621afcf21dd5
cc @staabm (cause I feel like you worked a lot on strict/loose comparison.