ratatool-sampling/src/main/scala/com/spotify/scio/Random.scala (3 lines):
	- line 64: // TODO: Is seed properly handled here
	- line 65: // TODO: is it necessary to setSeed for each instance like Spark does?
	- line 88: // TODO: is it necessary to setSeed for each instance like Spark does?


ratatool-sampling/src/main/scala/com/spotify/ratatool/samplers/util/SamplerSCollectionFunctions.scala (3 lines):
	- line 31: // TODO: What is a good number for tolerance
	- line 219: // TODO: Clean up magic number
	- line 259: // TODO: Clean up magic number


ratatool-sampling/src/main/scala/com/spotify/ratatool/samplers/BigSampler.scala (2 lines):
	- line 84: // TODO: for now leave it up to jit/compiler to optimize
	- line 99: // TODO: Rename --exact to something better


ratatool-sampling/src/main/scala/com/spotify/ratatool/samplers/BigSamplerBigQuery.scala (2 lines):
	- line 95: // TODO: Potentially reduce this and hashAvroField to a single function
	- line 143: // TODO: investigate if possible to move this logic to BQ itself


ratatool-sampling/src/main/java/org/apache/beam/sdk/io/gcp/bigquery/PatchedBigQueryTableRowIterator.java (1 line):
	- line 257: // TODO: This limitation is unfortunate. We need to give users a way to use BigQueryIO that does


ratatool-diffy/src/main/scala/com/spotify/ratatool/diffy/BigDiffy.scala (1 line):
	- line 784: // TODO: handle schema evolution