Bug: Strips away key information if path is recursive and only one key/value set

[kaelumania]

When I use the recurse version and there is only one value set in any of the child paths, e.g.

Diplomat::Kv.put('a/b/c', 'test')
Diplomat::Kv.get('a/b', {recurse: true}, :return, :return)

The the output is just

test

and I am missing the information from which childpath/key the value came from. This behaviour is implemented at https://github.com/WeAreFarmGeek/diplomat/blob/master/lib/diplomat/rest_client.rb#L79-L81

[johnhamelink]

Well spotted. The question is - should we detect if the path is recursive and return a different result or should we make it an explicit option that you set?

[johnhamelink]

I've decided I'm going to return the hash, but I will do this as part of a breaking change, which according to Semver will bump the version number up to 1.0.0. I'll do this after I've merged everything else that's non-breaking in and released 0.13.0.

[chaos95]

@johnhamelink According to Semver, major version 0 (that is, 0.y.z) is for initial development and does not guarantee any protection against breaking changes. You should feel free to break things between minor revisions when you're still in pre-release.

[johnhamelink]

@chaos95 Agreed, but considering the amount of people using this gem, we are now in release whether we like it or not 😄

[aaronbbrown]

Would be super sweet to see this get fixed, as I'm running into the same problem.

[EugenMayer]

Iam bumping this issue, since with 1.1.0 we have now convert_to_hash: true and this we can implement a good behavior here.

Right now, if you ask for test/this/out/key = 'test' with

Diplomat::Kv.get('test/', {convert_to_hash: true, recurse: true}, :return)

You get a plain parse exception that test cannot be "eached"

undefined method `each' for "test":String

@johnhamelink it seemed to be on the 1.0.0 roadmap but either it got forgotten or this issue is not the right home for my described bug. Could you tell me whether to create a new context for this?

For more cleanup to this ssue:
Pull request #51 should be closed since it is not based on the current hash-parser or was not aware back then in 2015, that it will be existing. I think, the patch for the issue above is far more trivial

[snoremac]

If it helps anyone, our fork changes behaviour such that single-valued results are packaged the same way as multi-valued ones.

https://github.com/caradvice/diplomat

[EugenMayer]

@snoremac wondering why you did not create a pull request? Thank you for your work!

[pklingem]

@snoremac thoughts on PRing?

[pierresouchay]

This has been added in 2.1.0