Implement TracerFactory

[jpkrohling]

Signed-off-by: Juraci Paixão Kröhling juraci@kroehling.de

Which problem is this PR solving?

• See Split the resolver from the "factory" opentracing-contrib/java-tracerresolver#30

Short description of the changes

• Implements the new TracerFactory interface, changing the current tracer resolver to get a tracer from this new class (so that we have only one place knowing how to create a tracer).

[codecov]

Codecov Report

Merging #527 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@             Coverage Diff              @@
##             master     #527      +/-   ##
============================================
+ Coverage     88.63%   88.64%   +0.01%     
- Complexity      515      517       +2     
============================================
  Files            66       67       +1     
  Lines          1891     1893       +2     
  Branches        243      243              
============================================
+ Hits           1676     1678       +2     
  Misses          141      141              
  Partials         74       74

Impacted Files	Coverage Δ	Complexity Δ
.../tracerresolver/internal/JaegerTracerResolver.java	100% <100%> (ø)	2 <1> (ø)	⬇️
...g/tracerresolver/internal/JaegerTracerFactory.java	100% <100%> (ø)	2 <2> (?)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8d325b9...6458d45. Read the comment docs.

[isaachier]

I'm glad I ended up using JaegerObjectFactory instead of JaegerTracerFactory in #509. :)

[jpkrohling]

codecov is lying: it claims that JaegerTracerResolver#resolve isn't being called, which is false. Running the tests in IntelliJ with code coverage, shows that it gets hit once:

I'm adding an explicit call to that method anyway, but codecov is lying!

[yurishkuro]

Afaik codecov just works off the coverage files we send to it. Worth checking if the execution is being captured there first.

[vprithvi]

@yurishkuro is correct here. You can inspect these files by clicking download on the codecov build view: https://codecov.io/gh/jaegertracing/jaeger-client-java/commit/c858eba8871b19269718525ab28b083e48446772/build

@jpkrohling Maybe I'm reading the coverage number wrong, but to me it seems like codecov shows that the line is covered

[jpkrohling]

It shows the line as covered now, as I'm making an explicit call to the method, but it wasn't before.

[jpkrohling]

I just tried locally with ./gradlew codeCoverageReport and there's something strange there...

If I have a code like this in the test:

    JaegerTracer tracer = ((JaegerTracerResolver) tracerResolver).resolve();
    assertNotNull(tracer);

I get a different coverage result:

I can't explain this, so, I'll just leave the test with the explicit call to make codecov happy. But it's still lying.

[isaachier]

I happen to agree that I've had my fair share of issues with codecov. Don't know if it stems from the language tooling or the codecov analysis.

[yurishkuro]

Still way better than coveralls

[jpkrohling]

Apart from the code coverage discussion, is this OK to get merged?