(Don't merge) Bump up to sbt 0.13.10-M1#210
Conversation
This is to test the quality of sbt 0.13.10 release candidates.
|
test run: https://scala-ci.typesafe.com/job/scala-2.11.x-integrate-community-build/211/consoleFull (link will 404 until worker comes online) |
could you "retest this please" please? |
|
peculiar. but, apparently just a transient error |
|
After waiting all day it has aborted :( https://scala-ci.typesafe.com/job/scala-2.11.x-integrate-community-build/212/consoleFull |
|
successfully built projects are cached, so progress made in an aborted run isn't lost. very likely this one will finish: https://scala-ci.typesafe.com/job/scala-2.11.x-integrate-community-build/213/consoleFull |
|
Oh excellent. Come on little 213 I believe. |
|
@eed3si9n Could you push the tag for 0.13.10-M1 please? I'm not sure what fixes are included in it. |
|
No SHA in the ivy metadata :( |
|
why M1 and not RC1? |
|
one more: https://scala-ci.typesafe.com/job/scala-2.11.x-integrate-community-build/214/consoleFull (404 til the worker spins up) |
|
RC1 was our first release candidate, but sbt/sbt#2411 was found. So M1 is our current, less certain, milestone release. I think it's probably sbt/sbt@e64b011 which is essentially sbt/sbt#2418 to fix sbt/sbt#2411. |
|
Oh and thanks for the rebuild. It's failed to download https://repo.typesafe.com/typesafe/ivy-releases/org.scala-sbt/interface/0.13.5/jars/interface.jar for some reason. Could you poke it again please? |
|
hmm, same failure - the exact error: |
|
sbt fails to find the compiler interface. The question is why does it try to get it in version 0.13.5? We changed the way the compiler interface and bridge are published with |
|
several subprojects ( here is a log of successful build of play2-core from a recent community build run using sbt 0.13.9: https://gist.github.com/SethTisue/2d9761ffcb68f3a6f05a. Note that the string
|
|
Who can be cc'd in from play that might be able to help here? |
|
@cunei could you help us with this? @SethTisue wrote:
|
|
trying it myself here: https://scala-ci.typesafe.com/job/scala-2.11.x-integrate-community-build/216/consoleFull (going slowly so far because of cascading rebuilds of unrelated, updated dependencies) |
|
hey, the play2-core build succeeded. so @eed3si9n, I think all you need to do is incorporate SethTisue@218687a into this PR |
|
I mean, I guess some possibility remains that the fact that this is now needed means there has been some weird regression in sbt. promising avenue of investigation, or rabbit hole with no rabbit at the bottom? I don't know. |
|
So play2-core was the only one out of 51 community build projects that failed to cleanly upgrade to 0.13.10-M1. Am I reading that right? (I expected a lot more projects to be honest) Thanks Seth. |
yup. I mean, there are entire huge categories of regressions this won't catch (those having to do with resolving binary dependencies, for example, or with publishing). but still, it's encouraging. |
|
I guess we can just close this, for now anyway. |
|
My recommendation would be to move to the Play 2.4.x branch - Play doesn't read its sbt version from a system property anymore, it uses whatever version of sbt is configured in build.properties. |
I believe that requires Java 8, so the 2.11.x(-jdk6) community build must stay on Play 2.3. The 2.11.x-jdk8 and 2.12.x community builds are already on Play 2.4. |
|
Shall I open "don't merge" pull requests against those to further test sbt 0.13.10-M1? |
|
sure! |
|
Ok, well it's not likely that we're going to fix this, as Play 2.3 is nearing end of life, and back porting the fixes we did in Play 2.4 to our build is a high effort high risk task that we don't have the time to do. |
no worries, it's easily worked around |
This is to test the quality of sbt 0.13.10 release candidates.
/cc @SethTisue