human parameter on the hot threads API returns a response that's not really human readable

[dedemorton]

When I run the hot threads API with ?human=true specified, the results are not really human readable, and specifying ?human=false does not seem to change the response.

In previous alpha versions, I believe that we returned a more readable response, as shown in the documentation for the hot threads API. I'm opening this issue in case this change was not intentional.

• Version: 5.0.0-alpha4
• Operating System: OS X
• Config File: NA
• Sample Data: NA
• Steps to Reproduce:
  Call the hot threads API and specify ?human=true. The call returns a response that seems to be in XML format, but (IMHO) isn't really human readable. Setting human=false returns the same document. This makes me wonder if the human parameter is working as expected.

[purbon]

@andrewvc something got wrong in your during your refactoring?

[nemonster]

I just saw the same thing running this command from the diagnostic. I used the exact syntax that was in the current alpha-4 documentation. Copied it.
/_node/hot_threads?human=true

What I get is unpretty json. It's essentially the same output as if you run it with no options at all.

Attaching a sample of what I got.
nodes_hot_threads.txt