Regression in sbt 1.9.5, likely zinc regression

[mdedetrich]

Reproduction steps

Scala version: N/A (see PR)

Updating an existing project to SBT 1.9.5, see apache/pekko#648

Problem

[error] /home/runner/work/incubator-pekko/incubator-pekko/stream/src/main/scala/org/apache/pekko/stream/impl/io/compression/GzipDecompressor.scala:29:52: Generated class org.apache.pekko.stream.impl.io.compression.createLogic$$anon differs only in case from org.apache.pekko.stream.impl.io.compression.createLogic$$anon (defined in DeflateDecompressor.scala).
[error]   Such classes will overwrite one another on case-insensitive filesystems.
[error]   override def createLogic(attr: Attributes) = new DecompressorParsingLogic {
[error]                                                    ^
[error] /home/runner/work/incubator-pekko/incubator-pekko/stream/src/main/scala/org/apache/pekko/stream/impl/io/compression/GzipDecompressor.scala:40:26: Generated class org.apache.pekko.stream.impl.io.compression.createLogic$$anon$inflating$ differs only in case from org.apache.pekko.stream.impl.io.compression.createLogic$$anon$inflating$ (defined in DeflateDecompressor.scala).
[error]   Such classes will overwrite one another on case-insensitive filesystems.
[error]     override case object inflating extends Inflate(false) with Step
[error]                          ^
[error] two errors found

Original notes

Still need to figure out what the root cause is, basically a project (Pekko) was updated to 1.9.5 and then MiMa started complaining about bincompat issues (see https://github.com/apache/incubator-pekko/actions/runs/6182223720/job/16781910666?pr=648). My initial suspicion/hunch is that it may be due to the new compiler bridge that was added in zinc which was brought in by the new sbt version (see scala/scala#10472)

[SethTisue]

I'm testing sbt 1.9.5 in the Scala 2 community build now — perhaps that will generate some insight.

fyi @lrytz

[mdedetrich]

So I doubt that this is due to the new compiler bridge because that was added for Scala 2.13 and MiMa is complaining about Scala 2.12 artifacts.

I am looking through the various changes put into Zinc recently (which is what new sbt 1.9.5 brings in) and the only one that comes to mind is sbt/zinc#1244 (it seems to be dealing with filenames which is what MiMa is complaining about).

The inliner was also enabled in Zinc for Scala 2.13, but I seriously doubt thats the issue (this would be some serious bug in the inliner and furthermore would only be applicable if Zinc on Scala 2.13 is used to compile Scala 2.12 artifacts which I don't think is the case)

[pjfanning]

The issue occurs with sbt +compile when you compile with latest https://github.com/apache/incubator-pekko (main branch) code but with project/build.properties edited to use sbt 1.9.5. You don't need to do a Mima check to reproduce it.

I checked and it only seems to happen with sbt ++2.12 compile - it does not happen with sbt ++2.13 compile

[SethTisue]

the only one that comes to mind is sbt/zinc#1244

@dwijnand wdyt, plausible culprit?

[He-Pin]

I tested with sbt 1.9.5 and 2.13.11, fails too.

[lrytz]

MiMa started complaining

Just to clarify - is it really MiMa complaining? The error message "... differs only in case from ..." is from the compiler (https://github.com/scala/scala/blob/v2.12.18/src/compiler/scala/tools/nsc/backend/jvm/PostProcessor.scala#L104-L105)

[mdedetrich]

Just to clarify - is it really MiMa complaining? The error message "... differs only in case from ..." is from the compiler (https://github.com/scala/scala/blob/v2.12.18/src/compiler/scala/tools/nsc/backend/jvm/PostProcessor.scala#L104-L105)

Your right, it happened in the MiMa step but it does seem to be the actual Scala compiler

[pjfanning]

@lrytz see #12868 (comment)

[lrytz]

Created scala-steward-org/scala-steward#3167 for now -- I don't know if this would work?

[lrytz]

I can confirm it's caused by sbt/zinc#1244. I'll continue digging, but probably won't have anything done during ScalaDays.

[mdedetrich]

I can confirm it's caused by sbt/zinc#1244. I'll continue digging, but probably won't have anything done during ScalaDays.

Thanks for diagnosing, enjoy the conference!

[SethTisue]

cc @eed3si9n