Adds Scalyr logging endpoint support

[mccurdyc]

Relevant (Referenced) Link(s)

• Fastly API Docs
• fastly/go-fastly

Validation

$ make test

git clone git@github.com:mccurdyc/cli.git /tmp/cli && \
(
    cd /tmp/cli && \
    git checkout mccurdyc/logging-scalyr && \
    make test && \
    rm -rf /tmp/cli \
)

[akappen]

We just released region support for Scalyr logging. The API documentation has been updated if you'd like to include support for that in the CLI. Thanks!

[pteichman]

Conflict fix is over at e556630

[mccurdyc]

@akappen Awesome! We will have to include in fastly/go-fastly and then cut a new release and then update the dependency version used in the CLI and then update this PR.

[mccurdyc]

@phamann I'm going to come back to this to address this comment, but I would like your thoughts on the CreateCommand.createInput() Create*Input pattern as I will adopt this in the other open PRs and come back to Scalyr once I cut a new go-fastly release.

[phamann]

👍 LGTM. Regardless of the small nits around certain input validations, I like the new crateInput() pattern overall and how you've tested it. I'm on the fence around the naming convention of scalyr_test.go. and api_test.go as this breaks convention from the rest of the project. What was your reasoning for keeping createInput() private and thus having this fork?

[phamann]

Whilst this is good, we don't validate negative numbers for version input anywhere else in the project. I'm also conscious of doing this kind of client-side validation when the API will do it anyway.

The big revelation for me here is why are the version fields in go-fastly uint's to avoid this completely 😱 (not your fault, just convention there I hadn't realised until now)

[mccurdyc]

Force-pushed rebasing master.

[mccurdyc]

Force-pushed squashing into a single commit.

[mccurdyc]

Force-pushed rebasing master.