I'm hesitant to do so. It would require PMD to bundle the kotlin compiler, which is dependent on the Kotlin version in use (experimental features may break even in minor versions, and there are no compatibility promises for an eventual Kotlin 2 release, as per https://kotlinlang.org/docs/reference/compatibility.html).

As far as I know, the Kotlin compiler, unlike the Java compiler, can't be told "this code is in version X", so supporting multiple Kotlin versions require bundling different versions of the compiler, which in turn means a really heavyweight PMD binary, and not doing so means forcing users to stick to an outdated PMD version or upgrade their codebases to whatever Kotlin version we decide to support.

Unless IntelliJ decides to join forces with us (I'd love to!), the same way Salesforce did for Apex / VisualForce support, and work with us to avoid these issues, I've some doubts this approach will be sustainable.

Using the Kotlin parser to get the UAST still needs mapping those nodes to our own Node structures for the pmd-core to be able to work on them. This is exactly what we do on Salesforce Apex code for instance, so the main difference is wether we use a grammar implementation from someone else, or our own. Both approaches have their benefits and their cons, I don't think one is clearly better over the other, so I'm ok with either on this particular aspect and it's a tradeoff I think we can work with.

On the other hand, we do have type and reference resolution in place for JVM-based languages. I know Kotlin can compile to other platforms, but probably most of our users for the time being are focused on Android support, so we are probably just fine were we stand at the moment.

Uh oh!

[kotlin] Add support for Kotlin #419

Description

Metadata

Metadata

Assignees

Labels

Type

Fields

Projects

Milestone

Relationships

Development

Issue actions