Commits on Feb 15, 2022

1. test: move output test to example, include zapr …

   To reproduce the loss of verbosity in
   klogr (kubernetes#294) we need a backend which
   actually emits the verbosity. zapr does that and also happens to be the logging
   implementation used in Kubernetes for JSON, so it makes sense to test with
   that.
   
   The expected output mapping may be useful also for testing in Kubernetes,
   therefore it gets exported by k8s.io/klog/v2/test.
   
   Because k8s.io/klog/v2 must not depend on zapr, the actual output test gets
   moved to the examples. The downside is that "go test ./..." no longer runs it.
   One has to remember to enter the "examples" directory. The GitHub action gets
   updated to do that.

   

   pohly committed Feb 15, 2022
   Copy the full SHA

   bb978b3 View commit details

   Browse the repository at this point in the history

2. klogr: pass verbosity level through to klog …

   That wasn't (and still isn't) necessary when klog itself emits the log entry
   because it doesn't log the verbosity. But when SetLogger was used to enabled
   logging through a backend logger, that logger may care about the verbosity
   level.

   

   pohly committed Feb 15, 2022
   Copy the full SHA

   a58a994 View commit details

   Browse the repository at this point in the history

3. github: split up test steps …

   Splitting up the tests ensures that each step runs in the right directory
   without having to rely on more complex shell scripting. That failures get
   reported separately is an added bonus.
   
   The simpler `(cd hack/tools && go test ./...)` wasn't supported by Windows.

   

   pohly committed Feb 15, 2022
   Copy the full SHA

   7dac793 View commit details

   Browse the repository at this point in the history