Thanks for the PR, much appreciated! For now, I pulled it into a new branch to get the API right.

It's not a huge problem to parse the encoding from the resource definition and hand it down to the Table factory method. Reading Resources from files should be easy, and the API for Tables and Datapackages would match. What's giving me headaches is whether to set an encoding on a URL-based Table in the first place. Usually, the web server should give us that.

In the context of Table, it should be fine to trust the web server and therefore not have an encoding param in the Table.fromSource method, but in the context of a Datapackage, there might be an encoding defined on a Resource which may or may not match the encoding of the individual URLs.

Essentially, it comes down to:

• File based Table:
  • no encoding -> use UTF-8.
  • Encoding given -> use it
• URL based Table:
  • no encoding -> use whatever the web server gives us OR use UTF-8 (?).
  • Encoding given: might clash with what the web server gives us if the Table API allows to specify it.

@roll How do other implementations handle this? or what's your opinion?

Hi, in Python we infer the encoding from a byte sample (buffer) if it's not provided. We tried to use an encoding from the HTTP headers but it's ofter missleading so we stopped using it.