Fix for non-null parentSpanId

[tcolgate]

No description provided.

[tcolgate]

I'm not sure on your preferred syntax here, but the parentId must be encoded if it exists.

[tcolgate]

At this point the failing tests seem to be in the stackdriver exporter (unrelated to this change)

[fabiogomessilva]

Hi @tcolgate,

Thanks for your contribution.

Actually, it is just a tslint issue.

lerna ERR! test Errored while running script in '@opencensus/exporter-jaeger'
...
lerna ERR! clang-format reported errors... run `gts fix` to address.

Please run 'gts fix' at package/opencensus-exporter-jaeger, run the tests again and commit the changes.

[tcolgate]

I might admit, I couldn't get the tests to run locally. They seem to fail, and I didn't see anything about gts in the output.

[fabiogomessilva]

Your last commit fixed the issue. I suppose you had managed to run the tests, right?

[tcolgate]

@fabiogomessilva no, but the gts fixed looked fine, I committed the result of that and let the CI tests run.
I'll try and get the tests working locally, but that may take me a while (node isn't my primary dev environment).

[kjin]

LGTM, thanks for the fix. At this point the CI failure seems to be related to degraded npm package download service in CircleCI. I'll re-trigger the run and see how it goes.

[tcolgate]

Is there a readme for setting up the dev environment?

[tcolgate]

@kjin I've added a test to check spans encode to the thrift spans. Seemed worth while, and forced me to get the local tests working.

[kjin]

@tcolgate Thanks for the extra test.

The hope is that setting up a dev environment is nothing more than npm install (with re-linking done between packages being lerna bootstrap). That probably is information that should go into a developer's guide, thanks for pointing that out. However, that should be the extent of "setting up" -- please feel free to file an issue if that is not working.

[kjin]

@tcolgate I just rebased since the Node 6 CI doesn't seem to want to work otherwise (still the same issue as I mentioned earlier, though it appears not to be affecting other PRs for some reason). If this passes I will land it.