remove path from failpoint

[lavacat]

See etcd-io/etcd#14727

[serathius]

This looks good for me. cc @ahrtr who had different opinion about the topic.
Can you fix the DCO check? Just run git checkout dd4869804163318da0745652c38501072f3c824d && git commit --amend --signoff && git push HEAD:remove-fp-path -f

[ahrtr]

cc @ahrtr who had different opinion about the topic.

Just as I mentioned in etcd-io/etcd#14727 (comment), it isn't a big deal. Since no feedback from other maintainers. I don't want to be blocked here. I agree to remove the pacakge path, although personally I tend to keep it.

[ahrtr]

We don't have e2e test for now, please make sure you manually verify the PR using examples

[serathius]

@lavacat Can you apply the suggestions by @ahrtr ?

[lavacat]

We don't have e2e test for now, please make sure you manually verify the PR using examples

Tried using examples.

./gofail enable examples
GOFAIL_FAILPOINTS="ExampleLabels=sleep(100)" go run examples/cmd/cmd.go

Also tried same fail point in 2 different packages to test warning.

None of the existing unit tests failed after the change, so I'm looking if I can add unit test.

[lavacat]

• added unit tests
• renamed failpath to name everywhere in runtime.go

@ahrtr FYI with your suggestion there was a deadlock, because register was calling Enable that acquires the same lock. Changed it to call enable. I think that was a typo in original code.

[lavacat]

• changed NewFailpoint/register logic to return existing failpoint if name is already registered
• updated readme
• minor unit tests improvements
  @serathius @ahrtr I think this is ready to merge. Please take a look.

[ahrtr]

Overall looks good to me with two comments:

1. Please fix the test failure;
2. Please consider using github.com/stretchr/testify in the unit test so simplify the implementation.

[lavacat]

Overall looks good to me with two comments:

1. Please fix the test failure;
2. Please consider using github.com/stretchr/testify in the unit test so simplify the implementation.

@ahrtr thanks, addressed both

[ahrtr]

LGTM

Thank you @lavacat