Release - April 2018 - Target RC date: 2018-04-02 - name: 0.13.0

[dslomov]

No description provided.

[dslomov]

@meteorcloudy lmk if you are not available on these dates

[lfpino]

#4928 may be a blocker for this release.

[dslomov]

We will be cutting RC approximately on April 7th /cc @philwo

[meteorcloudy]

I can do it by Friday.

…

On Thu, Apr 5, 2018, 3:03 AM Dmitry Lomov ***@***.***> wrote: We will be cutting RC approximately on April 7th /cc @philwo <https://github.com/philwo> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#4583 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AD-nthr1ulZYl_psAmzjY8KWWTS0x0Ptks5tlevngaJpZM4R5mvn> .

[meteorcloudy]

There are some jobs failing with Bazel@HEAD, but not with 0.11.0, we need to fix them

• bazel-skylib (mac)
• protobuf (mac)
• rules_nodejs (windows)
• rules_docker (all)
• rules_python (all)
• rules_typescript (all)

Ping the current Sheriff @vladmos

If they cannot be fixed in time, I will cut the release and cherry-pick the fixes later.

[meteorcloudy]

The mac failures are due to #4751, so I marked it as a release blocker.

[vladmos]

rules_python: there's an issue bazelbuild/rules_python#67 and a pull request bazelbuild/rules_python#83 that's supposed to fix it.
rules_typescript: red due to a remote caching problem, there's an existing issue for it: #4928 . Not sure if it's considered as a release blocker (@buchgr)
rules_docker: a pending issue: bazelbuild/rules_docker#363
rules_nodejs on windows: a pending issue: bazel-contrib/rules_nodejs#165

[meteorcloudy]

@vladmos Thanks Vlad for summarizing all these!
Seems like most of fixes are not on Bazel side, except #4751 and #4928 (possibly)
Then I'll cut a release based on yesterday's nightly, that is fdee70e

[meteorcloudy]

Release candidate 1 has been successfully cut, branch has been pushed (https://github.com/bazelbuild/bazel/tree/release-0.13.0)

[petemounce]

0.13.0-rc1 is available on chocolatey.

[laszlocsomor]

The release notes contains entries from the 0.12.0 release notes. I believe the release notes generator script is confused, because the last release was 0.11, not 0.12.

Also, there are duplicate entries, "None" entries, and some entries that were manually removed from the 0.12.0 release notes.

Please remove the following entries:

• python,runfiles: You can now depend on
  @bazel_tools//tools/runfiles:py-runfiles to get a
  platform-independent runfiles library for Python. See DocString
  of
  https://github.com/bazelbuild/bazel/blob/master/src/tools/runfiles
  /runfiles.py for usage information.
  ...
• The new "--local_tmp_root=" flag allows specifying the temp
  directory for locally executed actions.
  ...
• The new "--local_tmp_root=" flag allows specifying the temp
  directory for locally executed actions.
  ...
• None
  TESTED=presubmit, next cl
  ...
• None (select branches are mutually exclusive, so this is a no-op)

[jin]

Based off the nightly 0.13.0rc1 was cut from (https://buildkite.com/bazel/bazel-with-downstream-projects-bazel/builds/188), these are failing:

• rules_python 14.04, 16.04, macOS
• rules_typescript 14.04, 16.04
• rules_docker 14.04, 16.04, macOS
• bazel-toolchains 14.04 and 16.04: reported issue bazel-toolchains failing with 0.13.0rc1 #4992
• rules_nodejs on Windows: proposed a solution, pending fix rules_nodejs red on Bazel CI bazel-contrib/rules_nodejs#165
• Tensorflow 14.04, 16.04: android_binary.manifest_merger was removed, causing downstream breakages (cc @ahumesky). fixed tensorflow/tensorflow@a0410fb

/var/lib/buildkite-agent/builds/buildkite-ubuntu1604-dqml-1/bazel-downstream-projects/tensorflow/tensorflow/examples/android/BUILD:64:1: //tensorflow/examples/android:tensorflow_demo: no such attribute 'manifest_merger' in 'android_binary' rule

• Tensorflow Serving 14.04, 16.04: tracking issue for tf_serving to fetch upstream fix Update TF commit to fix tensorflow_serving with Bazel@HEAD tensorflow/serving#825

[meteorcloudy]

@laszlocsomor , thanks! I'll make sure they don't appear in the next RC.

[ahumesky]

I've sent out a change removing the usage of the manifest_merger attribute from tensorflow (It's basically an internal-only attribute and it was sort of an accident that it was released as part of the tensorflow demo)

[steeve]

Can we cherry-pick cb16c50 please pretty please ? Thank you!

[meteorcloudy]

@steeve cb16c50 is included in this release, no need to cherry-pick.

[steeve]

That's great thank you!

[brandjon]

FYI: Be advised that #5010 may lead to Skylark breakages if users are passing certain args by keyword. Skylark team expects the effect to be pretty minor, but if we're wrong we can make a patch. We're hoping to not have to do that since the behavior being broken is questionable anyway.

[laszlocsomor]

#5038 is a blocker.

[laszlocsomor]

Please cherry-pick f083e76 to fix #5038.

[meteorcloudy]

New cherry-pick fa36d2f for fixing #4928

[meteorcloudy]

0.13.0rc2 is created with scripts/release/release.sh create 0.13.0 fdee70e f083e76 fa36d2f , but haven't been pushed yet.
Building at https://buildkite.com/bazel/release/builds/30

[petemounce]

0.13.0-rc2 is up on chocolatey.

[jin]

Is it possible to cherrypick 4abd2ba to fix #5077?

[meteorcloudy]

Sorry @jin, #5077 isn't a regression in 0.13.0, so I think we'll have to wait until 0.14.0.
I'll try to release 0.13.0 ASAP!

[laszlocsomor]

#5079 is a blocker

[meteorcloudy]

Considering #4751 is also a bug already existing in 0.12.0, I'm removing it as a release blocker.
So, there is currently no unfixed release blocker. I'll wait for one more day in case there's any objection, if none, I'll release 0.13.0 on Wednesday.

[laszlocsomor]

Considering #4751 is also a bug already existing in 0.12.0

What about #5079? It's not a regression since 0.12.0, but it makes Bazel unusable for tests in the // package.

[meteorcloudy]

OK, then I'll cherry-pick the fixes for both #5079 and #5077. But still leaving #4751 not being release blocker, because we are not making progress on fixing it and the issue only happens when using http remote cache on macOS, which is not used often by our users. @laszlocsomor @jin

[laszlocsomor]

Please cherry-pick 2c95757 to fix #5079.

[meteorcloudy]

@laszlocsomor Thank you!

[jin]

@meteorcloudy thank you!

[meteorcloudy]

0.13.0rc3 has been created with
scripts/release/release.sh create 0.13.0 fdee70e f083e76 fa36d2f 4abd2ba 2c95757
Available at: https://releases.bazel.build/0.13.0/rc3/index.html
Thanks to @philwo !

[meteorcloudy]

A new release blocker is discovered: #5066

[meteorcloudy]

0.13.0rc4 has been created with
scripts/release/release.sh create 0.13.0 fdee70e f083e76 fa36d2f 4abd2ba 2c95757 f0cd2d2
Available at: https://releases.bazel.build/0.13.0/rc4/index.html

[meteorcloudy]

Unfortunately, rc4 is bad due to a merge conflict error, please don't use it. I'm fixing it.
Sorry for the trouble

[meteorcloudy]

Turned out #5066 is not a release blocker, we don't have to cherry-pick f0cd2d2

[meteorcloudy]

0.13.0rc5 has been created with
scripts/release/release.sh create 0.13.0 fdee70e f083e76 fa36d2f 4abd2ba 2c95757
Available at: https://releases.bazel.build/0.13.0/rc5/index.html

rc5 is for undoing rc4, it's exactly the same as rc3

[meteorcloudy]

We have the 0.13.0 release:
https://github.com/bazelbuild/bazel/releases/tag/0.13.0

[petemounce]

Chocolatey package is published.

[werkt]

Remote Ex for 0.13.0 is producing directories with unsorted file/directory lists. This was a regression introduced in 0.12.0 (verified with deb testing of 0.11.1, 0.12.0, and 0.13.0). Issue was filed as #5109 with supplied PR #5110. Recommend inclusion into a patch release, as this will be broken against any validating Remote Ex Execution Service (including buildfarm)

[werkt]

a729b9b introduced the remote ex regression

[meteorcloudy]

https://github.com/bazelbuild/bazel/releases/tag/0.13.1 Fixed the remote execution regression. 0.13.0 is done!