Avoid unnecessary calls to .toString in ViewData.checkAggregation

[steveniemitz]

Please answer these questions before submitting a bug report.

What version of OpenCensus are you using?

0.12.2, but present up to current version

What JVM are you using (java -version)?

java version "1.8.0_111"
Java(TM) SE Runtime Environment (build 1.8.0_111-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.111-b14, mixed mode)

I ran into this randomly running a CPU profiler on a live instance running OpenCensus 0.12.2. We use a custom exporter to export OpenCensus metrics into our own system. I noticed that a lot of CPU time was spent in ViewData.checkAggregation (https://github.com/census-instrumentation/opencensus-java/blob/master/api/src/main/java/io/opencensus/stats/ViewData.java#L252). Specific to our use case, we create Distribution objects with ~10,000 buckets to get high precision on our percentile calculations. createErrorMessageForAggregation is called for each View instance, regardless if the error message will be used or not. This then calls .toString on the Distribution instance, which calls toString on the bucket boundaries, which in my case was very large.

[sebright2]

I agree that we shouldn't generate the error message if it isn't needed. We should add a checkArgument method that takes an object that generates the message String lazily.

[songy23]

Thanks for investigating and reporting this issue. I'll send a PR to fix this soon.

[steveniemitz]

Thank you!