Convert function to property shouldn't insert explicit type if it was inferred previously #1011
Conversation
| elementsToShorten.add(property.typeReference!!) | ||
|
|
||
| if (!hasExplicitType) { | ||
| val linebreak = property.getter?.node?.treePrev |
There was a problem hiding this comment.
Ouch. This should be handled by a formatter rule, not by AST manipulations.
There was a problem hiding this comment.
I agree this isn't pretty. Currently, the formatter allows both, getters on the same line and on the next line. Forcing one of them could be to strict.
Maybe the KtPsiFactory could get an option to create a property with the getter on the same line. What do you think @yole?
Edit: Maybe that's already possible and we should just use the PsiFactory to create the whole thing instead of reconstructing the KtNamedFunction part by part?
There was a problem hiding this comment.
Actually yes, the way this intention replaces the internal child nodes of KtFunction is quite ugly. I think it would be much nicer to use KtPsiFactory.CallableBuilder to construct the property.
There was a problem hiding this comment.
I think the reason for the current way is because it preserves annotations, modifiers, receiver type and other things I probably forgot. If we redid it, we would have to manually take care of it.
cypressious
force-pushed
the
function_explicit_type
branch
from
37ca5b9
to
b21d55f
Compare
|
@yole I reworked the code and added some more tests. I think it's cleaner now. |
… to property shouldn't insert explicit type if it was inferred previously' Fixes #KT-14820
cypressious
force-pushed
the
function_explicit_type
branch
from
b21d55f
to
110d112
Compare
Fixes #KT-14820