Document thoughts on specification

[mgirlich]

@krlmlr I have written up some thoughts on the tibblify() specification. It would be great if you could have a look at it whether it makes sense to you and which conclusions we can draw from this.

Key notes:

• tib_*() specifies the input and the output format.
• tibblify() should be strict about the input format.
• I think tib_*() should capture the "idea" of the output format. Therefore, tib_col() should rather be an argument to tib_df().
• Overview of formats that should be possible to parse.
• Representation as a tibble

Key questions:

• Are there other input formats that should be supported?
• Does one of the tibble representations make sense?

Scope question

• The main motivation for tibblify was to be used with JSON resp. data resulting from parsing JSON. Can it also be used outside of this context? Would this influence the API

[krlmlr]

Thanks. I don't have enough mind space to do a deep dive now.

Do we have a comprehensive overview of the JSON and nested R list constructs that should be processible with tibblify? Should this be part of this documentation, could action items be deducted from that (e.g. #97)?