search

SUSE / container-suseconnect

Provides access to repositories inside containers using the host entitlements
Apache License 2.0
19 stars 16 forks source link

Cleanup and Documentation #44

Closed rhafer closed 4 years ago

rhafer commented 4 years ago

This adds documentation about the new public cloud related additions and cleans up a few minor things.

rhafer commented 4 years ago

Hm, not sure what it is yet. But it seems the TestReadConfigFromServerBadResponse is flaky. See: https://travis-ci.org/github/SUSE/container-suseconnect/jobs/700008966

Or maybe it's just travis that had a hickup.

mssola commented 4 years ago

Hm, not sure what it is yet. But it seems the TestReadConfigFromServerBadResponse is flaky. See: https://travis-ci.org/github/SUSE/container-suseconnect/jobs/700008966

That's weird. Anyways, it should be pretty easy to workaround. It looks like the error has a different message (even though they both refer to a parsing error). That might be about the Go version or something like that. I'd say that just checking that err is not nil https://github.com/SUSE/container-suseconnect/blob/master/internal/regionsrv/server_test.go#L112 should be enough.

rhafer commented 4 years ago

That might be about the Go version or something like that.

Hm, doesn't seem so. Then it would fail consistently on that go version. There were failed runs on go1.12 (and -tip). Also, looking at the encoding/json sources I think the errror we're expecting is the right one.

I'd say that just checking that err is not nil https://github.com/SUSE/container-suseconnect/blob/master/internal/regionsrv/server_test.go#L112 should be enough.

I'd rather not do that for now. We might be papering over some real errors with that. I now re-ran the tests quite a few times and they always succeeded. I think I'd blame the failures on a glitch in the travis workers for now.

Let's merge this PR as is and keep an eye on failing tests in the future.

rhafer commented 4 years ago

I'd rather not do that for now. We might be papering over some real errors with that. I now re-ran the tests quite a few times and they always succeeded. I think I'd blame the failures on a glitch in the travis workers for now.

Turns out that there actually is a real problem in the code see #45 for details.