Regression: false warning about non-used variable

[DavidPerezIngeniero]

This code snippet compiles with no warning in Scala 2.12.4:

    describe("procesos") {
      it("incluye los procesos") { f =>
        val modeloProceso= ModeloRdsBuilder.modeloProceso("p")  // <=== Warning reported here
        f.insertaModelo(modeloProceso)
        val proceso = EntidadDependiente(modeloProceso.entity, Option(123), "proceso", Seq.empty, 1)
        val modelo = ModeloRdsBuilder.modelo().copy(procesos = Seq(proceso))
        val md5Icono = "abc123"
        val ep = f.setUp(
          modelo = modelo // <========= Variable "modelo" is used here!!!!!!!!!!!!!!!!
        )
        f.insertaIcono(proceso.icono.get, md5Icono)
        val result = f.consultaWeb(ep).get
        val procesos = result.processes

        assert(procesos.length == 1, procesos)
        val found = procesos.head
        assert(found.id == proceso.entity.id)
        assert(found.description == proceso.titulo)
        assert(
          found.href == DataSetEditorFormReference(
            f.areaId,
            f.dataSetId,
            EditorDataSetId.procesoEditor(ep, 1)
          )
        )
        assert(
          found.icon == Option(IconReference(AreaId.AdminArea, IconSource.Rds, md5Icono))
        )
      }

but with Scala versions 2.12.5, 2.12.6, 2.12.7, and 2.12.8 it complains in this way:

VistaConsultaWebSpec.scala:494:13: local val modelo in value $anonfun is never used
         val modelo = ModeloRdsBuilder.modelo().copy(procesos = Seq(proceso))

I can provide privately full sources if necessary.
We compile with SBT 1.1.2, Java 8 and Linux 64 bits.
As we compile with the flag to treat warnings as errors, we cannot adopt latest Scala versions.

[SethTisue]

do you have a minimized, standalone reproducer?

[DavidPerezIngeniero]

No, I don't have now the time to invest on it.

[SethTisue]

@som-snytt does it look to you like something that might already be fixed in 2.13.x?

[som-snytt]

Is this ScalaTest? Is it a macro expansion? I don't know anything about ScalaTest.

On 2.12.5 there was scala/scala#6190 and follow-up scala/scala#6404 which had to do with for comprehensions, and similarly #11175 and #10287

I don't mind running a private build or looking at -Xprint:typer for this test. The other option is to turn off the warning for test.

[DavidPerezIngeniero]

That's right. It is a ScalaTest. We're using ScalaTest version 3.0.4.
Maybe some macro provided by ScalaTest isn't working well.

I'll try to upgrade to latest ScalaTest 3.0.5.

[DavidPerezIngeniero]

I've tested ScalaTest 3.0.5 and Scala 2.12.8 and I get the same warning.
Trying to modify slightly the source code in order to remove warning.

[som-snytt]

@DavidPerezIngeniero Thanks, I'll try to reproduce. I feel bad about changing source for a spurious warning. However, I like that ingeniero recalls the word ingenious.

[DavidPerezIngeniero]

I attach the full source file.
VistaConsultaWebSpec.scala.txt

[som-snytt]

I made a reduction but not a reproduction. I see f.setUp is also called in "construye el formulario del filtro correctamente", followed by asserts. Here is sample output of -Xprint:typer for my simplified test:

    VistaSpec.this.describe("procesos")(VistaSpec.this.it.apply("incluye los procesos").apply(((f: VistaSpec.this.FixtureParam) => {
      val modelo: String = "";
      val ep: VistaSpec.this.EditorPrincipalCW = f.setUp(modelo, f.setUp$default$2, f.setUp$default$3, f.setUp$default$4, f.setUp$default$5, f.setUp$default$6, f.setUp$default$7, f.setUp$default$8, f.setUp$default$9, f.setUp$default$10, f.setUp$default$11, f.setUp$default$12, f.setUp$default$13);
      val result: Int = ep.length();
      {
        val $org_scalatest_assert_macro_expr: org.scalactic.Bool = {
          val $org_scalatest_assert_macro_left: result.type = result;
          val $org_scalatest_assert_macro_right: Int = 1;
          org.scalactic.Bool.binaryMacroBool($org_scalatest_assert_macro_left, "==", $org_scalatest_assert_macro_right, $org_scalatest_assert_macro_left.==($org_scalatest_assert_macro_right), scalactic.this.Prettifier.default)
        };
        VistaSpec.this.assertionsHelper.macroAssert($org_scalatest_assert_macro_expr, ep, scalactic.this.Prettifier.default, org.scalactic.source.Position.apply("VistaSpec.scala", "Please set the environment variable SCALACTIC_FILL_FILE_PATHNAMES to yes at compile time to enable this feature.", 29))
      }
    }))

Because modelo is the first parameter, the default args are invoked in order. With filtro also specified, it becomes a block, so maybe that explains the difference in behavior:

      val ep: VistaSpec.this.EditorPrincipalCW = {
        <artifact> val x$1: String = modelo;
        <artifact> val x$2: String = filtro;
        <artifact> val x$3: VistaSpec.this.GridRds = f.setUp$default$2;
        <artifact> val x$4: Boolean = f.setUp$default$4;
        <artifact> val x$5: Seq[VistaSpec.this.Vinculo] @scala.reflect.internal.annotations.uncheckedBounds = f.setUp$default$5;
        <artifact> val x$6: Seq[VistaSpec.this.Privilegio] @scala.reflect.internal.annotations.uncheckedBounds = f.setUp$default$6;
        <artifact> val x$7: Int = f.setUp$default$7;
        <artifact> val x$8: String = f.setUp$default$8;
        <artifact> val x$9: Boolean = f.setUp$default$9;
        <artifact> val x$10: Seq[VistaSpec.this.ReferenciaLibro] @scala.reflect.internal.annotations.uncheckedBounds = f.setUp$default$10;
        <artifact> val x$11: Seq[VistaSpec.this.ProcesoFormulario] @scala.reflect.internal.annotations.uncheckedBounds = f.setUp$default$11;
        <artifact> val x$12: Seq[VistaSpec.this.Datum] @scala.reflect.internal.annotations.uncheckedBounds = f.setUp$default$12;
        <artifact> val x$13: VistaSpec.this.OcultaBotones = f.setUp$default$13;
        f.setUp(x$1, x$3, x$2, x$4, x$5, x$6, x$7, x$8, x$9, x$10, x$11, x$12, x$13)
      };

If you post the snippet of output with scalacOptions += "-Xprint:typer"for "incluye los procesos", I can look further. You might also try changing the setup call to f.setUp(modelo=modelo, filtro=null), in case that just makes it work for you. Later, I'll try to make my example code complex enough to trigger the issue, but I may not be successful.

[DavidPerezIngeniero]

Thanks for your work. I'll try shortly to provide more information.

[DavidPerezIngeniero]

Here is the requested data:

VistaConsultaWebSpec.txt

Using Scala 2.12.8.

[som-snytt]

Thanks. I spent some time with it, but unfortunately I see no clues. There are two previous calls to f.setUp(modelo); they are slightly different: one supplies a second arg, the other defines modelo outside the it block.

[DavidPerezIngeniero]

I'll ask the programmer who wrote the code to help me to find the clue.

[som-snytt]

Closing since we did not minimize, but I'll make an effort to check the latest fixes to warnings against the test frameworks.