gochomugo / elbow Goto Github PK

The setup hook is simply labelled setup. We should prefix elbow- for easier debugging of tests.

Allow replacing variables in strings in schema. For example,

(partial schema file)

{
    "params": {
        "username": "${USERNAME}"
    }
}

The variables can be provided through the options object, or from environment variables. If a variable is not found, default to an empty string and log a warning to console.

Having "status": 201 should test the http response's status code to be 201

We should be validating the schemas that we load before using them in the tests. This will avoid difficult-to-debug errors when the schema is ill-formatted.

The label used in it("label", function() {}); should be able to be customized or generated.

The current implementation reads all the schema files into memory, before executing the corresponding test for each of them. This will cause too much "buffering" (eat up memory resources) should the files be very many. We should instead use streaming to allow IO be more graceful.

Using a file listing the sequence in which requests should be fired would help avoid having to rename multiple files when we wish to insert a request, in between some other requests.

Should two or more schemas share the same endpoint, one of the schemas will overshadow the rest making them remain untested.

So why should two or more schemas share the same endpoint?

1. It might be possible that the Http response returned depends on the data passed

It would be extremely helpful to support other formats, such as JSON5, while writing the schema files.

We would have the option options.extensions for specifying which formats are supported. By default it would equal ["json"]. For example, if the user wanted to support JSON5, they would specify options.extensions as ["json", "json5"]. They would go ahead and add the necessary require extensions to allow elbow to do something like require("./sample.json5").

Allow passing headers, just like params, thus headers.