These are chat archives for typelevel/scala

10th
May 2017
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:41
@milessabin I was updating the project. Try again, without removing sbt-sensible
you only need to upgrade the scalaVersion
Miles Sabin
@milessabin
May 10 2017 10:42
Well, I guess the point I'm trying to make is that this seems to be an issue with sbt-sensible.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:49
I updated it after you tried it, you're giving feedback on an old version
Miles Sabin
@milessabin
May 10 2017 10:50
I don't understand ... I'm giving feedback on removing it altogether ... when I do and add an explicit scalatest dependency your project compiles.
If you're saying that if you leave it in, even after tweaking it, it still fails, then I still think its a problem with your plugin.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:53
I'm not going to remove the plugin. What changed in TLS vs LB scala that would cause inductive to behave differently?
Miles Sabin
@milessabin
May 10 2017 10:53
I'm not going to debug your plugin, so I guess you're just out of luck.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:54
I'm not asking you to debug the plugin, but it'd be useful to know what changed around inductives between TLS-3 and TLS-4
Miles Sabin
@milessabin
May 10 2017 10:55
As an aside, I'm afraid your own bug reports are nothing like up to the standards you demand for Ensime reports ... please follow your own advice.
Nothing changed.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:55
oh come on, I've given a repo and a command that fully reproduces the problem. If I got that, ENSIME would be laughing.
Miles Sabin
@milessabin
May 10 2017 10:56
Something in your plugin in forcing an explicit "org.scala-lang" dependency ... if you can find it and reproduce the problem in a build that doesn't use your plugin then I'll take a look.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:57
it's definitely being introduced by something sbt-sensible is doing, I'm trying to minimise
Miles Sabin
@milessabin
May 10 2017 10:57
Thank you.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 10:57
but it would be useful to know what changed in inductive so I know where to put my attention
Miles Sabin
@milessabin
May 10 2017 10:57
Nothing changed.
It's possible that you're seeing a difference due to the shift in baseline compiler version from 2.12.1 to 2.12.2 ... perhaps there's some logic in your plugin which hard codes that?
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:04
ok, minimised it
it's a bug in coursier I think
I bet you it's related to the dependencyOverride bug that you all don't think exists
Miles Sabin
@milessabin
May 10 2017 11:05
Is that the one you haven't reported?
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:05
btw, have you tried a fresh 2.12.2-bin-typelevel-4 project with a completely clean ivy folder, and seen what gets populated in there? I think you may be surprised
Miles Sabin
@milessabin
May 10 2017 11:08
That might be more useful reported as an issue on the coursier issue tracker.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:08
coursier/coursier#536
Miles Sabin
@milessabin
May 10 2017 11:09
:+1:
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:10
looks like the other issue with the dependency overrides is also a coursier bug
Miles Sabin
@milessabin
May 10 2017 11:11
Have you seen this: coursier/coursier#392
I believe that coursier has a typelevel option you need to specify.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:14
coursier/coursier#537
oh, but it was working before
the option is for the CLI launcher, not the sbt plugin by the looks of it
Miles Sabin
@milessabin
May 10 2017 11:17
I'm not seeing anything surprising in a freshly populated ivy cache ... what are you seeing?
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:19
try updateClassifiers
I was seeing a bunch of scala-lang jars
Miles Sabin
@milessabin
May 10 2017 11:21
I see the 2.10 jars for sbt. Nothing else.
Ghost
@ghost~540393fe163965c9bc2018ce
May 10 2017 11:25
hmm, maybe it's more of the same coursier bug
hopefully Alex can get to the bottom of it
probably it was working only because the versions matched up
Torsten Schmits
@tek
May 10 2017 13:31
so this error is most likely due to some dep pulling in a scala-lang artifact, yes?
Miles Sabin
@milessabin
May 10 2017 14:49
Yes, but I think you'd have to be trying pretty hard to do that ... the changes I made to SBT in the 0.13.13 timeframe should be causing all those deps to be rewritten into org.typelevel.
It's possible that this is a coursier problem, as @fommil suggests.
Torsten Schmits
@tek
May 10 2017 17:21
kind-projector is pulling in scala-lang…do I have to specify CrossVersion.patch for it?
Torsten Schmits
@tek
May 10 2017 18:57
getting close!
Alan Johnson
@acjay
May 10 2017 19:01
i'm getting an error knownDirectSubclasses of X observed before subclass Y registered, but it doesn't identify what line of code is causing this
Alan Johnson
@acjay
May 10 2017 19:13
it's rather confusing, because i'm pretty sure i'm definining my classes before spray-json-shapeless is invoked, which is the only thing i know of in my project that relies on knownDirectSubclasses
Alan Johnson
@acjay
May 10 2017 20:52
haha, back to scalac it is, then :). was hoping to take advantage of faster typeclass derivation
Alan Johnson
@acjay
May 10 2017 21:07
blob
Github bummer :(
Alan Johnson
@acjay
May 10 2017 21:17
i have no idea what i'm talking about when it comes to working on the compiler, but it appears there's something called an attachement, which i'm guessing propogates data through the compilation process, and my error appears to be surfaced because of the presence of the KnownDirectSubclassesCalled attachement. where the attachement is created, i'm wondering if it's possible to attach the context of what line invoked knownDirectSubclasses?