Reports 128-bit ids to Zipkin

[codefromthecrypt]

This allows Wingtips to report 128-bit trace IDs to Zipkin, by parsing
this high bits of a 32-character hex-encoded ID.

See openzipkin/b3-propagation#6

[codefromthecrypt]

Note: this doesn't initiate 128-bit traces. This might be desired at some point. For example Brave has the following builder to do that. Note span ids in a trace are still 64-bit.

Brave.Builder.traceId128Bit(true)

[codecov-io]

Current coverage is 98.53% (diff: 81.81%)

Merging #34 into master will decrease coverage by 0.27%

@@             master        #34   diff @@
==========================================
  Files            12         12          
  Lines           674        682     +8   
  Methods           0          0          
  Messages          0          0          
  Branches        121        124     +3   
==========================================
+ Hits            666        672     +6   
- Misses            2          3     +1   
- Partials          6          7     +1   

Powered by Codecov. Last update 9fa505f...94c2328

[nicmunroe]

I don't think this line will ever be reached - the test data is setup so that it should blow up on the previous line.

[codefromthecrypt]

ah right! ok I'll duplicate the test then? or is there a way to add another dimension to it (i wonder)

[nicmunroe]

@adriancole whatever you want to do is fine - feel free to duplicate the test, or add another dimension. Either way. If you want to add another dimension and you're not familiar with the com.tngtech.java:junit-dataprovider all you should have to do is separate the new dimension with a | pipe, something like:

@DataProvider(value = {
        "                                      |    true", // less than 16 chars
        "                                      |    false", // less than 16 chars
        "123e4567-e89b-12d3-a456-426655440000  |    true", // longer than 32 chars
        "123e4567-e89b-12d3-a456-426655440000  |    false", // longer than 32 chars
etc etc

And then add the new dimension to the method args: public void huge_long_method_name(final String badHexString, boolean goodNameForNewDimension).

[codefromthecrypt]

thanks.. I rejigged the one test to show the parsing exceptions were the same regardless of offset. Then, made one that looks redundant, but I think useful to have in the context of where this is all put together.

[nicmunroe]

Looks great. I'll go through at some point and fill in the remaining code coverage - it'll be good for me to make sure I understand 100% what you added. And I added a ticket for optionally supporting 128 bit trace ID initiation. Thank you!

[codefromthecrypt]

thanks for the support!