Move SamplingStatus to spi package

[isaachier]

Signed-off-by: Isaac Hier ihier@uber.com

Which problem is this PR solving?

• Resolves Move SamplingStatus out of internal package #501.

Short description of the changes

• Moves SamplingStatus from internal package to spi package.

[codecov]

Codecov Report

Merging #502 into master will decrease coverage by 1.55%.
The diff coverage is n/a.

@@             Coverage Diff             @@
##             master    #502      +/-   ##
===========================================
- Coverage     89.76%   88.2%   -1.56%     
+ Complexity      563     498      -65     
===========================================
  Files            69      65       -4     
  Lines          2061    1848     -213     
  Branches        261     239      -22     
===========================================
- Hits           1850    1630     -220     
- Misses          132     141       +9     
+ Partials         79      77       -2

Impacted Files	Coverage Δ	Complexity Δ
...main/java/io/jaegertracing/spi/SamplingStatus.java	66.66% <ø> (ø)	2 <0> (?)
...racing/internal/samplers/ProbabilisticSampler.java	81.81% <ø> (ø)	12 <0> (ø)	⬇️
...internal/samplers/GuaranteedThroughputSampler.java	92.3% <ø> (ø)	7 <0> (ø)	⬇️
...ing/internal/samplers/RemoteControlledSampler.java	89.53% <ø> (-1.72%)	17 <0> (+1)
...tracing/internal/samplers/RateLimitingSampler.java	66.66% <ø> (ø)	6 <0> (ø)	⬇️
...tracing/internal/samplers/PerOperationSampler.java	93.47% <ø> (ø)	18 <0> (ø)	⬇️
.../jaegertracing/internal/samplers/ConstSampler.java	66.66% <ø> (ø)	4 <0> (ø)	⬇️
...n/java/io/jaegertracing/internal/JaegerTracer.java	89.83% <ø> (+0.86%)	24 <0> (-2)	⬇️
...rtracing/internal/reporters/CompositeReporter.java	71.42% <0%> (-28.58%)	5% <0%> (-2%)
...java/io/jaegertracing/zipkin/ZipkinV2Reporter.java	71.42% <0%> (-28.58%)	2% <0%> (-1%)
... and 23 more

---

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 3947cbf...961c74b. Read the comment docs.

[isaachier]

@yurishkuro @jpkrohling @vprithvi any thoughts about this change?

[yurishkuro]

Long term I think this is the right change. Short term I'd prefer not to do it, because the whole point of package renaming was to give us the opportunity to revisit & clean up the interfaces. In particular, returning the status objects is not very efficient as it requires memory allocation. Also, Won and I discussed that we may want to change the sampler signature to be more efficient, in particular by accepting the current time stamp (used by the rate limiter), or the whole span that already contains the time stamp, the span id, and the operation name (but passing the span creates a dependency loop of sorts).

So I would've actually preferred not having the Sampler interface in the spi package just yet.

Why not making this change blocks you from implementing a Sampler? The internal package is still accessible.

[isaachier]

I don't believe I need this at the moment. Just checking in. Should I close this or leave it open just in case we want this change later on?