Add RBAC CLI targets for goreleaser/makefile - #136
Conversation
…t/cloud, -cloud, etc.
|
(If current state gets approved I'll add some dummy rbac command etc. so semaphore passes) |
|
seems reasonable to me |
Codecov Report
@@ Coverage Diff @@
## master #136 +/- ##
=========================================
- Coverage 35.14% 35.1% -0.04%
=========================================
Files 30 30
Lines 1781 1783 +2
=========================================
Hits 626 626
- Misses 1087 1089 +2
Partials 68 68
Continue to review full report at Codecov.
|
Cody A. Ray (codyaray)
left a comment
Cody A. Ray (codyaray)
left a comment
There was a problem hiding this comment.
since we're building/shipping two binaries for CP 5.3 next release, this approach makes sense
| # TODO: [CLI-92] we delete the semaphore cache during release to workaround an issue with semaphore and goreleaser | ||
| - rm -rf $GOPATH/pkg/mod | ||
|
|
||
| # NOTE: This will put all builds into the same ./dist folder. There is no way to configure goreleaser output directories per-build, only per-project. |
There was a problem hiding this comment.
is there a TODO here or is this already accounted for in the makefile?
There was a problem hiding this comment.
I was told CI is using the .goreleaser.yml file while locally I know we're building with the -mac.yml or -linux.yml files. So, CI just throws everything in ./dist, whereas local builds shove stuff into ./dist/ccloud or ./dist/confluent. But I think that's fine since CI isn't pushing to s3, we're doing that manually with make targets (which use subfolders correctly)?
There was a problem hiding this comment.
goreleaser will be pushing binaries to s3... but yeah, we push the archives ourselves from the Makefile.
The other approach is to have two separate goreleaser-CI files (one per binary). Then we can disable S3 altogether for confluent.
But I think we want binaries in S3 for both of them to support self-update. (I'm going to start saying "self-update" instead of auto-update... since we give users the choice. And maybe this won't be as scary sounding to management/product folks here. :)
Cody A. Ray (codyaray)
left a comment
Cody A. Ray (codyaray)
left a comment
There was a problem hiding this comment.
lgtm
we can do more packaging stuff as a follow-up
…#136) * add current user to the statement property as statement-ownder' * recover .go-version * recovered order of imports * simplified 'if'
* chore: stop swallowing error when LSP client fails to be initialized (#136) * flink statement resume * flink statement update * fix tests * list compute pool flag first * updated with revised plan * fix tests * rebase * rebase test working * Revert Last change * Formatiing * Failing test case * Test Cases fix * Lint * Lint test case --------- Co-authored-by: Yannick Pferr <37211050+YannickPferr@users.noreply.github.com> Co-authored-by: Brian Strauch <bstrauch@confluent.io> Co-authored-by: Tushar Malik <tmalik@confluent.io>
* chore: stop swallowing error when LSP client fails to be initialized (#136) * flink statement resume * flink statement update * fix tests * list compute pool flag first * updated with revised plan * fix tests * rebase * rebase test working * Revert Last change * Formatiing * Failing test case * Test Cases fix * Lint * Lint test case * review --------- Co-authored-by: Yannick Pferr <37211050+YannickPferr@users.noreply.github.com> Co-authored-by: Brian Strauch <bstrauch@confluent.io> Co-authored-by: Tushar Malik <tmalik@confluent.io>
3 participants
Builds now go to
dist/ccloudordist/confluentinstead of justdist.Definitely feels like duplicating stuff here, but goreleaser is limited in its configurability (e.g. can't have one project that outputs binaries to different directories, sigh).
There's no
dist-confluenttarget since that isn't fleshed out yet.