Currently, we keep the state of analysis in memory and pass it over along the pipelines, and use it as part of the input to compute the next build number to rerun the test.
However, it's better to compute the next build number directly from the data points from all previous reruns.
This will make the code cleaner, and easier to test.
Comment 1 by chanli@chromium.org
, Dec 6 2016Owner: chanli@chromium.org
Status: Started (was: Available)