eugene yokota

Here's the full stacktrace: ``` [error] last tree to typer: Ident(x) [error] tree position: line 283 of /private/tmp/xenomorph/modules/core/src/main/scala/xenomorph/SchemaF.scala [error] tree tpe: Boolean [error] symbol: value x [error] symbol definition: x:...

https://github.com/sbt/zinc/blob/52286d0862cde407596efc3bb9ba6259d365d2ab/internal/compiler-bridge/src/main/scala/xsbt/Dependency.scala#L260 https://github.com/sbt/zinc/blob/52286d0862cde407596efc3bb9ba6259d365d2ab/internal/compiler-bridge/src/main/scala/xsbt/Dependency.scala#L346 https://github.com/sbt/zinc/blob/52286d0862cde407596efc3bb9ba6259d365d2ab/internal/compiler-bridge/src/main/scala/xsbt/Dependency.scala#L407-L410

Could you come up with a minimized-ish example, so we can use it as a test? The fix itself should be fairly simple, which I'd be happy to work on.

For reference, here's an example script - https://github.com/sbt/zinc/tree/develop/zinc/src/sbt-test/source-dependencies/abstract-type

Zinc is used on a clean build as well.

Interesting. That would give better usability compared to invalidating all sources :)

Could we deprecate all inlining?

http://slides.com/cb372/inline-specialized-berlin-2016#/27 says it's slightly slower when used with HotSpot, in that particular benchmark.

For now, we should at least print warnings if it's anything less than 100% compilation. To potential contributor. Related source might be https://github.com/sbt/zinc/blob/v1.1.7/internal/zinc-core/src/main/scala/sbt/internal/inc/IncrementalCommon.scala#L125-L132

Let's start the petition for the call-site inline removal. Where do we start?