Add POST codable route support without input object.

[mbarnach]

Add a new codable route for POST with no object as input.

Description

This allows to create a resource without arguments, like this:

router.post("/resource") { (respondWith: (Resource?, RequestError?) -> Void) in
  respondWith(resource, nil)
}

Motivation and Context

It is not possible to use the POST verb without argument, making it weird to create a new resource from an endpoint that has no dependencies.

How Has This Been Tested?

External testing done. I'm unsure about where to add the test if needed.

Checklist:

• If applicable, I have updated the documentation accordingly.
• If applicable, I have added tests to cover my changes.

[ianpartridge]

Thanks for the PR! We'll get it reviewed.

[djones6]

I'm not sure whether REST has an opinion on doing this, but I can imagine there may be circumstances where you might want a resource that contains only values generated by the server. Could you elaborate a little on the use-cases for this?

Regarding tests, you can add a new test to TestCodableRouter, copying the example from testBasicPost: https://github.com/IBM-Swift/Kitura/blob/master/Tests/KituraTests/TestCodableRouter.swift#L143

I think it would also be useful to add a test to the SwaggerGenerator to make sure the generated document makes sense - unfortunately the tests here are not quite so well crafted, but you can follow the example of an existing POST route test here:

• route handler: https://github.com/IBM-Swift/Kitura/blob/master/Tests/KituraTests/TestSwaggerGeneration.swift#L123
• route registration: https://github.com/IBM-Swift/Kitura/blob/master/Tests/KituraTests/TestSwaggerGeneration.swift#L214
• path assertion: https://github.com/IBM-Swift/Kitura/blob/master/Tests/KituraTests/TestSwaggerGeneration.swift#L229
• content assertions: https://github.com/IBM-Swift/Kitura/blob/master/Tests/KituraTests/TestSwaggerGeneration.swift#L620

[mbarnach]

I've updated the tests.
For the use case, I guess it could be seen as a semantically different way of saying GET /resource but with creation, like for a token, a new "element", etc.
For the REST opinion, I've found that (interesting) thread: w3 lists. Long story short, not so common, but can make sense (so an option to have?). There is also some valid arguments in that StackOverflow thread, especially about the caching involved eventually.

[djones6]

This line generates a compile warning because new is not used. Could change to let _ = or maybe there's a neater way to check the type?

[mbarnach]

Fixed! Thanks.

[djones6]

@ianpartridge The code for this change looks okay, what do you think?

[djones6]

The Travis failure on this job is unrelated to this PR, and has been happening on the CRON builds for master since about 3 weeks ago, eg: https://travis-ci.org/IBM-Swift/Kitura/jobs/648208574
Timeouts in tests when running with the combination of KITURA_NIO on OSX. @harish1992 could you take a look?

[CLAassistant]

All committers have signed the CLA.