Release 0.20 - November 2018

[laurentlb]

@philwo Can you manage the 0.20 release?

Target RC date: November 5th

[philwo]

@laurentlb Yes, I can do this.

[dslomov]

#6319 is a release blocker.

[dslomov]

I suggest I co-own this release to try out Communicating Breaking Changes and Managing Breaking Changes

[dslomov]

This baseline looks good to me: 7bf7f03.

(https://buildkite.com/bazel/bazel-with-downstream-projects-bazel/builds/558#322c009a-8d9d-4474-ad60-f6408b3377ff)

[dslomov]

rc1 is out:

scripts/release/release.sh create 0.20 7bf7f03

release notes created with help of #6639

[petemounce]

Published to chocolatey

On Fri, 9 Nov 2018, 11:38 Dmitry Lomov, ***@***.***> wrote: rc1 is out: scripts/release/release.sh create 0.20 7bf7f03 release notes created with help of #6639 <#6639> — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#6117 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEQWI_C_scjdE_qv-UvYTIm6H8E_9xRks5utWk5gaJpZM4WhlP4> .

--

…

-- Sent from my phone. Please excuse typos and brevity, but never text speak.

[dslomov]

First cherrypick: fd52341

[dslomov]

rc2:

./scripts/release/release.sh create --force_rc=2 7bf7f03 fd52341

[dslomov]

Actually
./scripts/release/release.sh create --force_rc=2 0.20 7bf7f03 fd52341

[petemounce]

Published rc2.

[ola-rozenfeld]

Is it too late to set baa1786 as the baseline? It (and the prerequisite fbfb3cb) introduce an --incompatible_remote_symlinks behavior, required for remote API V2 implementation.
If it is too late, can we please cherrypick both of these? Thank you!

[dslomov]

Is it too late to set baa1786 as the baseline? It (and the prerequisite fbfb3cb) introduce an --incompatible_remote_symlinks behavior, required for remote API V2 implementation.
If it is too late, can we please cherrypick both of these? Thank you!

No, we are not cherrypicking features.
Also, does --incompatible_remote_symlinks have an issue associated with it? (per policy)

[dslomov]

#6675 is a release blocker

[dslomov]

rc3
./scripts/release/release.sh create --force_rc=3 0.20.0 7bf7f03 fd52341 241f28d

[ola-rozenfeld]

Yes, #6631.
Okay, I see. Dang it, I thought I'd make it. No worries, next one.

[petemounce]

rc3 is on chocolatey.

[scentini]

@dslomov can you please cherrypick f7e5aef?
It was comitted and cherrypicked into 0.19.2 to fix users using --crosstool_top=@bazel_tools//tools/cpp:default-toolchain

[dslomov]

Spounds like #6662 is a release blocker

[dslomov]

rc4:

./scripts/release/release.sh create --force_rc=4 0.20.0 7bf7f03 fd52341 241f28d f7e5aef

[dslomov]

Before and after cherry-picking f7e5aef bazel-toolchains pass. Sounds like an AI for #6662 post-mortem - insufficient test coverage? /cc @scentini @katre

[meteorcloudy]

@dslomov Please cherry-pick d2920e3 to fix #6731

[dslomov]

I will cherry-pick #6731 fix as a 'serious bug' exception to release policy (as far as we know the original commit didn't actually fix any issue and was speculative)

[dslomov]

rc5: ./scripts/release/release.sh create --force_rc=5 0.20.0 7bf7f03 fd52341 241f28d f7e5aef d2920e3

[dslomov]

rc5 promoted to release

[philwo]

@dslomov We discovered a Bazel bug on Windows that, in combination with a performance improvement that I did to our shell test framework, is now causing CI jobs to fail. It's technically not a regression in Bazel 0.20.0, because the bug existed before.

We have a fix for that here that we could cherry-pick into a patch release: bdf7a6b

Or we could accept this as a known issue in 0.20.0 and try to develop a workaround (would probably look like a selective rollback of the performance improvement just on Windows) just for our own test framework.

Which would you prefer?

FYI @meteorcloudy

[philwo]

Update: I found a possible workaround and sent it to Yun for code review.