Yun Peng

Results 201 comments of Yun Peng

> I'm hoping the answer is "flip both and adjust whatever breakages we see." We should do it in the reverse order to make sure downstream is green, otherwise we'll...

The result looks good. You can ignore the rules_nodejs failure. But except that, --incompatible_config_setting_private_default_visibility didn't break any downstream project, so you can flip it. --incompatible_enforce_config_setting_visibility is breaking Envoy and TensorFlow,...

6.0 is currently scheduled to be cut on Sep 26, see https://github.com/bazelbuild/bazel/issues/16159 But there are still many release blockers, see https://github.com/bazelbuild/bazel/issues?q=is%3Aopen+is%3Aissue+milestone%3A%226.0.0+release+blockers%22, and it's likely the cut day will be postponed...

Bumping to P1 since it's blocking 6.0 release

Flipping this flag is breaking downstream projects: https://buildkite.com/bazel/bazel-at-head-plus-downstream/builds/2678#_ @gregestren Please check https://buildkite.com/bazel/bazelisk-plus-incompatible-flags/builds/1297 before flipping the flag I'm rolling back the flag flip now

@gregestren Can you please make sure the incompatible pipeline says this flag is ready to flip before actually flipping it? This is causing a very red downstream: https://buildkite.com/bazel/bazel-at-head-plus-downstream/builds/2717#_ We'll have...

Re rules_pkg version in Bazel: https://github.com/bazelbuild/bazel/blob/master/distdir_deps.bzl#L300-L309 /cc @aiuto Looks like we just had 0.8.0 release yesterday

> Greg: I suppose you want to backport that to the 6.0 RC branch. Maybe not? If we are not cherry picking the flip of this flag in 6.0.

https://buildkite.com/bazel/bazelisk-plus-incompatible-flags runs nightly, please add "migration-ready" label when you want the pipeline to test this incompatible flag.

/cc @c-parsons @oquenchil Can you look into this? This is blocking 6.0 cut since it's breaking TensorFlow. My guess is https://github.com/bazelbuild/bazel/commit/dadc49e437018f482640ed76fae5307daf9911a8 somehow caused the command line to be too long?...