Documentation does not make it clear how to specify keys for Literal storage

[qbolec]

Afer reading https://www.npmjs.com/package/nconf I was surprised that

nconf.argv().env().defaults({
  'port':8815,
  'faye:endpoint': "http://localhost:8813/pubsub",
})

does not work. I finally guessed that the correct way to do that is:

nconf.argv().env().defaults({
  'port':8815,
  'faye':{
    'endpoint': "http://localhost:8813/pubsub",
  }
})

but this is not clear from examples given on the mentioned site, nor from wording.

More importantly it's not clear to me what is the semantic, of the later form - how are two or more such literals merged together? For example if I have

  'faye':{
    'password': "lolcat",
  }

somewhere in the hierarchy with a higher priority (say in overrides()), then, what will be the result of get('faye:endpoint') - will it be undefined, or "http://localhost:8813/pubsub"? My experiments show that the objects will be merged. The question is important for cases where merging values from different places together does not result in a semantically good value (a particular example I have in my mind is defining a list of servers for db sharding, where merging 2 production servers with 1 development server is not what you want). This is why I expected that the syntax for Literal storage is to use 'faye:endpoint' as the key - this way it would be explicit what is my intention: if I wanted to propose a default for the whole faye branch of the config tree I would write:

{
  'faye':{
    'endpoint':  "http://localhost:8813/pubsub",
  }
}

and if I only wanted to add a possibly missing child faye:endpoint I'd write:

{
  'faye:endpoint':"http://localhost:8813/pubsub",
}

[tekgator]

Was trapped in the same problem, thanks for the detailed description. I agree it should be as you described. For now I can at least continue with my coding.