This message was deleted.

Hi Ben, This topic will be a focus for the 8.x development line where Gradle intends to make improvements to Kotlin DSL build script compilation performance. 8.0 already starts with improvements, more to come. Currently, for raw script compilation performance, Groovy DSL is still ahead. However this impacts mostly very large projects when doing build script changes. For regular development use cases - working on project code - caching (regular and configuration one) levels the playing field. The main benefit of the Kotlin DSL is the way it helps discovery of Gradle types and extensions, when used with IDE support. If you are learning Gradle or discovering some plugins, go ahead and use it.

Or for helpful error messages when you mess up the syntax. 🙂

@Louis Jacomet thanks for the info. from what i gather, with 7.x, there's no good reason to switch over to kotlin for performance. i will keep a look at the 8.x changes.

My hunch there is that Kotlin will never be faster than Groovy. Best case, it's maybe going to be in the same range once K2 ships and Kotlin compiles faster. But the Kotlin compiler does more stuff (type inference, type checking and what not) so it can't really beat Groovy. So as others said, I wouldn't switch just for performance but there are other very compelling reasons to switch (autocomplete, type safety, ...)

That's correct, the Groovy compiler is so permissive that it'll be very difficult to be as fast. For example GroovyC doesn't even do compilation warnings, it simply cannot.

I always wonder about that last part actually. How come this is not possible with Groovy? Run the Groovy compiler and turn build.gradle into a .jar that is stored remotely?

It is possible, it's just not implemented (yet?)

Gotcha 👍

Same for compilation avoidance

Dang! Is this because the 1.8 compiler is slower?

why abi change became slower? 🤔 Groovy became faster as well. Kotlin DSL looks little faster as initial measurement

Comparing build duration against previous measurements doesn´t make sense. Comparing the ratio does.

I wonder if script compilation could reuse new IC from Kotlin/JVM for abi changes 🤔

The gap is not just the Kotlin compiler performance change. There are also some changes due to other areas of Gradle at play. But there definitely was a performance regression with 1.8 in our isolated tests. We are discussing this with the compiler team and will try to mitigate.