deps: update abseil and grpc dependencies#16687
Conversation
Signed-off-by: Antonio Vicente <avd@google.com>
repokitteh-read-only
bot
added
the
deps
|
Blocked by grpc/grpc#26359 and possibly grpc doing a release unless we chose to apply a local patch. Or merging in #16029 before updating the grpc dependency. |
Signed-off-by: Antonio Vicente <avd@google.com>
|
Windows failure looks like a clang compiler crash. Not sure how to get the input files that it wants attached to the bug report: https://dev.azure.com/cncf/4684fb3d-0389-4e0b-8251-221942316e06/_apis/build/builds/76968/logs/71 /retest |
|
Retrying Azure Pipelines: |
|
Preference from a dependency point of view is to not introduce a patch. Experience shows that patches will make future maintenance of the dependency more difficult / fragile /cc @htuch |
|
Bother, the windows clang compiler crash seems to be deterministic. |
Upgrading the grpc dependency is important for testability of the ALPN extension. We can try to ask grpc of a release once they accept a patch to their build under C++ 17 and see how that goes. It would be good to track dependencies more closely, we fell behind on grpc by 6 months and multiple releases. |
We have a log warning system that runs on dep check i think. Can we turn that into an issue filer on postsubmit? |
That would be great. Since you have more context on what is available, could you file an issue to track that improvement? |
|
@antoniovicente I'm working to get an updated Once I get the checker update merged I'll work on bumping the deps. |
|
With @envoyproxy/dependency-shepherds hat on, if it's going to be fixed in a reasonable timely way upstream, let's just do that and skip patch; i.e. if it's going to happen in a week or so. We don't need to consume point release versions, we often use a SHA from main branch in this situation temporarily until the next point release. If this is not going to happen upstream, then we can merge this as long as a tracking bug is filed to remove the patch. Patches are expedient, but in the longer term they slow down upgrades and make it even harder to keep up with releases. |
Signed-off-by: Antonio Vicente <avd@google.com>
Signed-off-by: Antonio Vicente <avd@google.com>
Signed-off-by: Antonio Vicente <avd@google.com>
Signed-off-by: Antonio Vicente <avd@google.com>
|
grpc/grpc#26359 was merged 4 days ago. No point release includes it yet. I guess that me being OOO the past week worked in the favor of skipping the patch. Except that even newer grpc also needs a newer version of abseil, so I'm also updating that. Hopefully picking up 6 months of changes to grpc and abseil won't cause any problems. |
antoniovicente
changed the title
bazel/repository_locations.bzl
Outdated
| urls = ["https://github.com/abseil/abseil-cpp/archive/{version}.tar.gz"], | ||
| use_category = ["dataplane_core", "controlplane"], | ||
| release_date = "2020-11-24", | ||
| release_date = "2021-06-07", |
There was a problem hiding this comment.
UTC for this commit is 2021-06-03 and this is why CI has failed. As per prior conversation it would be preferable if we could use the latest release of abseil at https://github.com/abseil/abseil-cpp/releases/tag/20210324.2. There is a parallel PR to bump abseil at #16681
There was a problem hiding this comment.
I see comments about #16681 having trouble compiling under Windows. Given all these difficulties, I would strongly prefer to go with a patch to grpc and decouple the abseil version bump from the grpc version bump.
Can I please go back to the version with the patch?
|
/retest #16907 for the test that failed in the most recent attempt. |
|
Retrying Azure Pipelines: |
repokitteh-read-only
bot
removed
the
deps
|
So it looks like latest absl works fine with CL. |
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/retest |
|
Retrying Azure Pipelines: |
Signed-off-by: Antonio Vicente <avd@google.com>
repokitteh-read-only
bot
added
the
deps
Signed-off-by: Antonio Vicente <avd@google.com>
- Adding new enum to source/client/stream_decoder.cc to reflect changes in envoyproxy/envoy#17044 - Adding a binding for external dependency on gtest introduced in envoyproxy/envoy#16687 - Sync to latest envoy included changes to `.bazelrc` and `ci/run_envoy_docker.sh` copied directly. Signed-off-by: William Juan <66322422+wjuan-AFK@users.noreply.github.com>
Signed-off-by: Antonio Vicente <avd@google.com>
5 participants
No description provided.