Descriptor definition; clarify JSON/YAML

[roll]

• fixes Yaml as well as JSON for Data Package descriptor files datapackage#292
• fixes Naming convention of additional properties in the descriptor datapackage#663

---

It's another take on frictionlessdata/datapackage#40. This pull request:

• removes duplication using the common definitions section
• keeps strictness regarding JSON requirements as well as datapackage.json naming

Although technically, it doesn't matter how datapackage.json file is named (anyway referenced by URI) it might be nice to keep naming strictness as it's already in v1 and it provides some benefits e.g. regarding usage analysis (for example, searching on github for datapackage.json etc)

[khusmann]

When available as a file a descriptor MUST be named datapackage.json and it MUST be placed in the top-level directory (relative to any other resources provided as part of the data package).

Nice. Another place where this strictness is appreciated is when datapackages are zipped for distribution. Implementations can be pointed to a datapackage zip and selectively extract resources from it, without the user needing to unzip into a folder. I think this is helpful for adoption by novice users... it allows us to to a drop-in replacement for an *.xlsx file (which is also just a zip of tabular data).

I describe this use case in more detail here: frictionlessdata/frictionless-r#193

[cloudflare-workers-and-pages]

Deploying datapackage with    Cloudflare Pages

Latest commit:	7fb1809
Status:	⚡️  Build in progress...

View logs

[roll]

ACCEPTED by WG (6/9)