Spurious errors when recompiling

[Alexey Romanov]

This looks like a possible scalac bug, but I'd expect people to have
run into it before if that was the case.

After checking out https://github.com/alexeyr/scala-time, I can
compile it once (using `mvn compile`). However, touching
src/main/scala/org/scala_tools/time/Implicits.scala and running `mvn
compile` again produces 56 errors similar to

[ERROR] F:\MyProgramming\scala-time\src\main\scala\org\scala_tools\time\StaticDateTime.scala:29:
error: value second is not a member of Int
[ERROR] def nextSecond = now + 1.second

For some reason, they don't use the implicits defined in
Implicits.scala, even though they can access it (there is no

object Implicits is not a member of package org.scala_tools.time

error). This can be reproduced with SBT:

sbt clean; sbt compile;
change Implicits.scala enough to produce different classes (e.g. add a
line "object Implicits2 extends Implicits")
sbt compile

But to make things even weirder, SBT produces a competely different
set of 24 errors!

Rampion at Stack Overflow was also able to reproduce it using scalac
directly: http://stackoverflow.com/a/9964884/9204

So, is it a bug? And if so, where: scalac or SBT and scala-maven-plugin?

Yours, Alexey Romanov

[Dave]

I can reproduce this with sbt 0.11.2 (24 errors) and mvn 3.0.2 (56
errors)
The work around is indeed to do a clean after changing a source file

I couldn't reproduce it with scalac or fsc. I could edit the source
file and compile without clean.
So I guess it only happens with a build tool.

On 1 apr, 20:42, Alexey Romanov <alexey.v.roma...@gmail.com> wrote:
> This looks like a possible scalac bug, but I'd expect people to have
> run into it before if that was the case.
>

> After checking outhttps://github.com/alexeyr/scala-time, I can

> compile it once (using `mvn compile`). However, touching
> src/main/scala/org/scala_tools/time/Implicits.scala and running `mvn
> compile` again produces 56 errors similar to
>

> [ERROR] F:\MyProgramming\scala-time\src\main\scala\org\scala_tools\time\StaticDateT­ime.scala:29:

[Alexey Romanov]

See https://issues.scala-lang.org/browse/SI-5639 for a reproduction with scalac.

Yours, Alexey Romanov

[Dave]

Yes indeed. Now I see it. I don't even have to edit and it is still in
scala-2.10.0-M2.
Looks like the compiler forgets to put the implicits in scope in the
second run.

Strange though that there is an edit needed for scala-time for sbt/mvn
and that direct compilation with scalac/fsc just works. So I think
that the latter bug is slightly different but related.


C:\scala-2.10.0-M2\examples\implicitsbug>scalac *.scala

C:\scala-2.10.0-M2\examples\implicitsbug>scalac *.scala
Bar.scala:6: error: type mismatch;
found : Int(1)
required: pack.age.Baz
def baz : Baz = 1
^
one error found


On 2 apr, 11:58, Alexey Romanov <alexey.v.roma...@gmail.com> wrote:
> Seehttps://issues.scala-lang.org/browse/SI-5639for a reproduction with scalac.

> >> Yours, Alexey Romanov- Tekst uit oorspronkelijk bericht niet weergeven -
>
> - Tekst uit oorspronkelijk bericht weergeven -

[Dave]

And it seems only a bug for top level imports.

This below works over and over again:

package pack.age

object Quux {
import pack.age.Implicits._

def baz : Baz = 1
}


On 2 apr, 12:51, Dave <dave.mahabiers...@hotmail.com> wrote:
> Yes indeed. Now I see it. I don't even have to edit and it is still in
> scala-2.10.0-M2.
> Looks like the compiler forgets to put the implicits in scope in the
> second run.
>
> Strange though that there is an edit needed for scala-time for sbt/mvn
> and that direct compilation with scalac/fsc just works. So I think
> that the latter bug is slightly different but related.
>
> C:\scala-2.10.0-M2\examples\implicitsbug>scalac *.scala
>
> C:\scala-2.10.0-M2\examples\implicitsbug>scalac *.scala
> Bar.scala:6: error: type mismatch;
>  found   : Int(1)
>  required: pack.age.Baz
>         def baz : Baz   = 1
>                           ^
> one error found
>
> On 2 apr, 11:58, Alexey Romanov <alexey.v.roma...@gmail.com> wrote:
>
>
>

> > Seehttps://issues.scala-lang.org/browse/SI-5639fora reproduction with scalac.

> > - Tekst uit oorspronkelijk bericht weergeven -- Tekst uit oorspronkelijk bericht niet weergeven -

[Johannes Rudolph]

Would you add this information to the bug tracker as well, perhaps
with a link to the discussions on the mailing list not everyone is
tracking scala-user.

Thanks for creating the isolated test case, this is great.

Johannes

--
Johannes

-----------------------------------------------
Johannes Rudolph
http://virtual-void.net

[Alexey Romanov]

I think edit is needed with mvn/sbt only to get them to recompile the
files :) They simply don't compile if they already have up-to-date
class files and if for some reason they choose different files to
recompile after editing that would explain different sets of errors.

Yours, Alexey Romanov

[Dave]

It's only strange that direct compiling of scala-time with scalac and
fsc doesn't have these errors.

C:\Users\Dave\scala-time>c:\scala-2.9.1-1\bin\scalac.bat -classpath "C:
\Users\Dave\.ivy2\cache\org.scala-lang\scala-library\jars\scala-
library-2.9.1-1.jar;C:\Users\Dave\.ivy2\cache\joda-time\joda-time\jars
\joda-time-2.1.jar;C:\Users\Dave\.ivy2\cache\org.joda\joda-convert\jars
\joda-convert-1.2.jar" ./src/main/scala/org/scala_tools/time/*.scala -
d ./target -verbose

compiles without errors over and over again (no matter if I edit or
not)

On 2 apr, 14:32, Alexey Romanov <alexey.v.roma...@gmail.com> wrote:
> I think edit is needed with mvn/sbt only to get them to recompile the
> files :) They simply don't compile if they already have up-to-date
> class files and if for some reason they choose different files to
> recompile after editing that would explain different sets of errors.
>
> Yours, Alexey Romanov
>
>
>
> On Mon, Apr 2, 2012 at 14:51, Dave <dave.mahabiers...@hotmail.com> wrote:
> > Yes indeed. Now I see it. I don't even have to edit and it is still in
> > scala-2.10.0-M2.
> > Looks like the compiler forgets to put the implicits in scope in the
> > second run.
>
> > Strange though that there is an edit needed for scala-time for sbt/mvn
> > and that direct compilation with scalac/fsc just works. So I think
> > that the latter bug is slightly different but related.
>
> > C:\scala-2.10.0-M2\examples\implicitsbug>scalac *.scala
>
> > C:\scala-2.10.0-M2\examples\implicitsbug>scalac *.scala
> > Bar.scala:6: error: type mismatch;
> >  found   : Int(1)
> >  required: pack.age.Baz
> >        def baz : Baz   = 1
> >                          ^
> > one error found
>
> > On 2 apr, 11:58, Alexey Romanov <alexey.v.roma...@gmail.com> wrote:

> >> Seehttps://issues.scala-lang.org/browse/SI-5639fora reproduction with scalac.

> >> - Tekst uit oorspronkelijk bericht weergeven -- Tekst uit oorspronkelijk bericht niet weergeven -

[Alexey Romanov]

Yes, that does look as if there are slightly different bugs.

Yours, Alexey Romanov