Create and send codecov report#34900
Conversation
|
I'm going to move all the jenkins entrypoint changes to #34911 and rebase this on that branch. |
0ca00c4 to
d8a216d
Compare
|
Can we have two test-unit modes, one with coverage and one without? |
33a4a3b to
f6912b5
Compare
|
Sounds good, I restored the old unit test script, so we can run without coverage. |
|
build is green! coverage report is at https://codecov.io/gh/moby/moby/commit/f6912b553d7e9f3bf8d11c0218bc0678a9ee216d |
|
I guess the github integration isn't working yet? Since we don't get a message from codecov on this PR. It may be that the codecov-bash isn't capable of detecting the PR number on jenkins. I'll look into it. |
|
@dnephin I believe I create the webhook, can you retrigger / repush ? |
f6912b5 to
8ff99e3
Compare
|
build is green, but no report :( |
|
Although there was a banner on https://codecov.io/github/moby/moby/commit/8ff99e35693c5e5a08924baac8b06414e6691881 saying "build is not done yet", and now that banner is gone. So codecov does know that the build finished, it just didn't send a report. Maybe the config has to be in master first? |
client/client_test.go
Outdated
There was a problem hiding this comment.
Why do we need a dependency to mark a skipped test?
There was a problem hiding this comment.
We don't, but it's better this way. Instead of reporting some meaningless message this prints the condition that returned true, so the literal runtime.GOOS == "windows". Additional messages are also supported, but generally unnecessary.
There was a problem hiding this comment.
This is a waste of dependency space. The skip message is sufficient.
There was a problem hiding this comment.
I disagree. Also, this dependency is not being added in this PR, so it's not really relevant.
|
Does this actually work? On most projects we have running codecov, the numbers are completely wrong. |
|
It doesn't work yet, but the number in the report https://codecov.io/github/moby/moby/commit/8ff99e35693c5e5a08924baac8b06414e6691881 generally seem plausible. Which numbers are completely wrong? Coverage percentage? |
Usually, the differential percentage. The local coverage tools are very good, but they tend not to take into account how Go does coverage and it skews the numbers. |
|
What's the status on this? |
|
Codecov is setup and running, but something is not configured properly, because this PR doesn't show the codecov status. I'll rebase and see what happens. It might be that the config needs to be in master? |
Signed-off-by: Daniel Nephin <dnephin@docker.com>
Because we merge master into the branch before running tests, so the actual git sha does not exist on any git remote. Signed-off-by: Daniel Nephin <dnephin@docker.com>
8ff99e3 to
b2faf24
Compare
|
The report is here: https://codecov.io/github/moby/moby/commit/b2faf24925ea0f820db62e1f3d96ee347972f12a (still pending while CI is still running). |
|
Edit: I spoke too soon, that variable is already set. |
|
Still LGTM for me 👼 |
|
LGTM 2 |
Branched from #34911
Create a code coverage report while running unit tests, and upload it to codecov.
This TestNewEnvClient was failing locally so I added some doc strings to debug the failures.