Bump json-schema/JSON-Schema-Test-Suite to 1.2.0 + some tests improvements

[mirfilip]

Things done:

• Simplify tests directory structure. Namespace is still unchanged.
• Change drafts tests that we read from json schema test suite to be an associative array where a key is a file.json / suite description / test case description

Failing tests output before the change:

1) JsonSchema\Tests\Drafts\Draft4Test::testValidCases with data set #75 ('{"bar":1,"baz":2}', '{"properties":{"foo":{"not":{}}}}')

Failing tests output after the change:

1) JsonSchema\Tests\Drafts\Draft4Test::testValidCases with data set "not.json / forbidden property / property absent" ('{"bar":1,"baz":2}', '{"properties":{"foo":{"not":{}}}}')

It uses description fields from suite and test case. This may seem an overkill, but it really simplifies hunting for the test to fix when you have a readable path.

• Update json-schema/JSON-Schema-Test-Suite to 1.2.0. Pay attention here, important:
  If we really want to state which version of the draft the validator follows, we need to compare against the newest available version. I presume everyone agrees ;) So, 1.2.0 bring three failure specs:
  • format.json / Draft 3 / optional
  • format.json / Draft 4 / optional
  • not.json / Draft 4 / required

Sad thing, there is no mechanism to skip one test case from not.json. For the time being, I've skipped the whole file as it's now my priority to fix that case as it's required by spec. Still, I think it doesn't hurt us as I doubt people are going to dig tests to see the compliance with spec (I'm working on documenting that).

/cc @bighappyface

[bighappyface]

+1

I looked into this in the past and stuffed it into my ever-growing to-do list... Thanks @mirfilip for bringing us up to the latest in terms of both the test site and a more sane PSR-4 directory structure.

[mirfilip]

@bighappyface Thx for the check. This will inevitably cause some PRs to need of rebase but I figured there are not too many PRs so it's a good time for such change. I'm thinking of doing the same for src, what do you think?

Also, do you want to have every PR squashed to a single commit?

[bighappyface]

@mirfilip I think squashed PRs are for the best. We need to add a CONTRIBUTING file to help clarify our expectations.

Yes, I say do the same for src while you are at it, and thanks ;)

[mirfilip]

@bighappyface squashed. This is the final version. I will tackle src structure separately.

[jojo1981]

+1

[bighappyface]

+1

I'm going to let this one sit in master and bump at a later time.