Add support for datapackages

[roll]

Overview

Current task configuration/descriptor doesn't support datapackages because of structure:

files:
settings:

Ideas

We could introduce attribute datapackages (singular or plural?) wich will be mutually exclusive(?) with files:

datapackages:
settings:

If there will be more than 1 datapackages it will require a few tasks inside one build (based on current goodtables capabilities).

Tasks

• [ ] add datapackages key to goodtables.yml
• [x] implement support on goodtables.py level - https://github.com/frictionlessdata/goodtables-py/issues/153
• [ ] tests

[roll]

@pwalsh @danfowler Is multiple datapackages per repository our use case?

[pwalsh]

@roll sure, why not?

[roll]

@pwalsh just to be sure (cc @amercader)

[pwalsh]

one or many data packages also ultimately represents some sort of file tree, and therefore can be reduced to the files: type of interface we use here anyway.

I might be good, as you said, to have:

datapackages:
files:
settings:

All possible as part of goodtables.yml (and not mutually exclusive). Whether this is then all just reduced to a single list of target files, or is a set of different jobs, is mostly just an implementation detail, with various pros and cons either way I suppose.

[amercader]

MVP if time, if not defer

[roll]

requires https://github.com/frictionlessdata/goodtables-py/issues/153

[roll]

@danfowler @amercader So Dan's preparing pilots with datapackage.json but this feature is not even for Alpha (mid of January). Is it correct?

[pwalsh]

@amercader assigned to you for scoping and estimates.

[amercader]

@roll can you add a mini spec and an estimate? You had a plan for this IIRC

[roll]

@amercader done :+1: