Upgrade to Bazel 0.12.0

[pauldraper]

No description provided.

[pauldraper]

Failing on test_scala_library_expect_failure_on_missing_direct_java. Looks like we're losing the label.

[ittaiz]

This is strange. @tomlu @cushon any chance you can take a look? I think this is related to the SJD changes though the odd thing here is that AFAICT we are creating the ijar here when we compile java files. https://github.com/bazelbuild/rules_scala/blob/master/scala/scala.bzl#L321

[cushon]

ijars created by java_common.compile do not currently get stamped, and we can't fix that until the label is passed in to the API. This is partially what is being discussed in https://groups.google.com/forum/#!topic/bazel-discuss/mt2llfwzmac.

[ittaiz]

Right... Long and complicated thread. Thanks for the clarification!

…

On Sat, 14 Apr 2018 at 1:11 Liam Miller-Cushon ***@***.***> wrote: ijars created by java_common.compile do not currently get stamped, and we can't fix that until the label is passed in to the API. This is partially what is being discussed in https://groups.google.com/forum/#!topic/bazel-discuss/mt2llfwzmac. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#481 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABUIF7K12j5RBcn-2NwUwWwGFbKTuvm2ks5toSKYgaJpZM4TTAsl> .

[johnynek]

looks like the test for strict deps needs to be updated. It failed.

[ittaiz]

We’re finalizing the behavior. I’ll update here tomorrow or Thursday the latest. How does that sound?

…

On Tue, 24 Apr 2018 at 4:35 P. Oscar Boykin ***@***.***> wrote: looks like the test for strict deps needs to be updated. It failed. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#481 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABUIF3S-Tz5fHbohwhhh2V1sCi_Ox2aJks5troFNgaJpZM4TTAsl> .

[ittaiz]

ok, the situation here is a bit complicated.
Bazel 0.12.0 indeed broke strict-deps add_dep support for code that doesn't use ijar (for example if you use the current scala_import or the filegroup from maven_jar)
internally we've worked around it by having a script iterate the log and take various hints to fix the suggestions but this is heavily based on internal assumptions.
The solution for most issues (except for exports) will probably come from: https://docs.google.com/document/d/1ubah6phuvWnugShtVgSQnaopQ1BtKtNxQASVwGZA7k0/edit?usp=sharing

I think the pragamatic approach here is to loosen the assertion in the test so that it covers strict-deps error but not the add_deps message until the above is implemented and supported in rules_scala.
This will allow to move forward with rules_scala's CI versions.

WDYT?

cc @natansil @or-shachar for thoughts as well

[natansil]

@ittaiz I'm fine with loosening the tests. it's just the java ones right?
I suppose an issue should be opened that states the add_deps message tests should be re-instated once the solution is implemented in bazel itself.

[johnynek]

Note, we are using the rules with bazel 0.12 at Stripe. No issues that I know of yet.

[jjudd]

Lucid has also been running on 0.12 without issue. Strict/unused dep recommendations did break for us, so we changed all of our java_imports to scala_imports and are running off the work done in this branch: #487