Publish Android artifacts on Maven #744
Conversation
eoger
force-pushed
the
maven-publishing
branch
27 times, most recently
from
0c058eb to
7c9983d
Compare
|
Note: we should probably make a github release from this branch (with a different version number) before merging. |
Yeah, agreed 👍 |
| @@ -393,7 +421,7 @@ def with_dockerfile(self, dockerfile): | |||
| "servobrowser/taskcluster-bootstrap:image-builder@sha256:" \ | |||
| "0a7d012ce444d62ffb9e7f06f0c52fedc24b68c2060711b313263367f7272d9d" | |||
| ) | |||
| .find_or_create("docker-image." + digest) | |||
| .find_or_create("appservices-docker-image." + digest) | |||
There was a problem hiding this comment.
This will regenerate the Docker image. I assume you've seen this work in TC recently?
| @@ -298,6 +324,8 @@ def build_worker_payload(self): | |||
| deindent("\n".join(self.scripts)) | |||
| ], | |||
| } | |||
| if len(self.artifacts) > 0 and "chainOfTrust" not in self.features: | |||
There was a problem hiding this comment.
This seems wrong. There are some tasks that shouldn't have the CoT enabled because they're not trustworthy or don't want to trust their inputs. Wouldn't this mean that any push to a PR, for example, will have CoT enabled?
| else: | ||
| task_title = "{} - Build and test".format(module) | ||
| task = android_task(task_title, libs_tasks) | ||
| task = android_task("{} - Build and test".format(module), libs_tasks) |
There was a problem hiding this comment.
Why not include some blurb about publishing for release builds? It's cheap and might just save somebody some thought. It also ties into my concern about CoT that I raise below. Is everything a release build now? That doesn't match my model of how CoT should work.
| @@ -282,6 +287,14 @@ def build_worker_payload(self): | |||
| "artifact_id": self.artifact_id, | |||
| } | |||
|
|
|||
| # XXX: Beetmover jobs that transfer the `forUnitTests` maven.zip need | |||
There was a problem hiding this comment.
Hmm. This really tells me that we can't exploit the richness of Maven. I wish this was more clear; as a consumer of m.m.o, it's weird to me that you're not "Just" publishing my ZIPed directory. That is, I could have POMs, BOMs, classifiers, extensions... and m.m.o would completely fail. I don't really know what the action is here: I've never seen meaningful docs for m.m.o, so even documenting wouldn't actually help a consumer like me.
eoger
force-pushed
the
maven-publishing
branch
2 times, most recently
from
82a60fe to
e228d99
Compare
eoger
force-pushed
the
maven-publishing
branch
4 times, most recently
from
9606186 to
0a49320
Compare
|
Blocked on #1071 |
|
Blocked on mozilla-releng/scriptworker#332 |
| @@ -111,8 +110,11 @@ import com.sun.jna.Platform as DefaultPlatform | |||
| ext.libsGitSha = "git --git-dir=${rootProject.rootDir}/.git diff --name-only master -- :/libs".execute().text.allWhitespace ? | |||
| "git --git-dir=${rootProject.rootDir}/.git rev-parse HEAD:libs".execute().text.trim() : null | |||
|
|
|||
| // Use in-tree libs from the source directory in CI or if the git SHA is unset; otherwise use | |||
There was a problem hiding this comment.
Might be worth using System.getenv('CI') in the ext.libsGitSha computation, maybe.
Although, arguably it wouldn't be a bad thing for CI to use downloaded libs if the libs dir hasn't changed... Although, it probably would need a slightly different command to determine that than the above.
eoger
force-pushed
the
maven-publishing
branch
3 times, most recently
from
4c43af1 to
1a8882a
Compare
Co-authored-by: Mihai Tabara <[email protected]>
Fixes #252