Improve compile-error feedback messages

[monoti]

Summary

When I run sbt stryker, compilation fails due to it tries to mutate either >= or <= operation to ==.
The correct syntax of Spark's condition for equal method is ===.

I have prepared a repo to reproduce the issue: stryker-spark-example.

It will throw below error when you run sbt stryker.

[error] /Users/stryker-spark-example/target/stryker4s-572897741400037258/src/main/scala/SimpleDFHelper.scala:27:32: value or is not a member of Boolean
[error]         df.where($"col1" == 10 or col("col2") <= "B")
[error]                                ^
[error] /Users/stryker-spark-example/target/stryker4s-572897741400037258/src/main/scala/SimpleDFHelper.scala:35:47: type mismatch;
[error]  found   : Boolean
[error]  required: org.apache.spark.sql.Column
[error]         df.where($"col1" >= 10 or col("col2") == "B")
[error]                                               ^
[error] two errors found
[error] (Compile / compileIncremental) Compilation failed

Stryker4s config

stryker4s {
  mutate: [ "src/main/scala/*.scala"]
}

Stryker4s environment

stryker4s 0.14.0

Your Environment

software	version(s)
Scala version	2.13.6
Spark version	3.2.0
Build tool & version	1.3.13
Operating System	macOS Big Sur

Note that the issue occurs on Scala 2.12.x and Spark 2.4.x as well.

[hugo-vrijswijk]

Hi, does Stryker4s still continue? Since 0.14.0 Stryker4s is able to deal with compile errors and mark them in the report. We don't have enough type information to know Spark uses ===, but it should still be able to continue

[monoti]

Hi @hugo-vrijswijk,
Yes, it can continue. Thanks. Is there any plans to make Stryker mutation support Spark?

[hugo-vrijswijk]

@monoti I don't think that will happen anytime in the near future. Otherwise if we ever do type analysis for every mutant we'd have to check a possible mutation to both == and === which, in Scala, can come from a lot of different places.

The mutant is marked as a compile error, but I think the sbt plugin should have better output so it doesn't show this as an error (it's not, it's just part of the process of removing non-compiling mutants)