I had this concern when the change was made to make the bisect post results to the dashboard.

The information is still available under steps called 'Post bisect result' that are done after every revision and at the end, and the argument was made that the responsibility of formatting the information for human consumption belonged with the dashboard.

My opinion is that at least for now it would be more helpful to declutter the output of the bisect by nesting steps more logically, remove unnecessary repetition, etc.

+Prasad had good ideas about how the end result should look like.

Other related issues:

- I find it really hard to figure out where is the stdio log of the benchmarks as they were run during the bisect steps. For example, often the error: "Metric ['foo', 'bar'] was not found in the test output." means that the browser crashed or somehow stopped prematurely while running the benchmark. But I can't figure out the exact reason without looking at the stdio.

- The old bisect script had some "partial bisect results" which were useful to look at even if the bisect job eventually went on to fail due to some other random reasons; or to make a call on a possible suspect CL even if it's taking a lot longer to finish the full bisect. Not sure if we have a similar functionality on the new bisect recipe, or where to find that output.

 Issue 606533  has been merged into this issue.

Design doc for this feature: https://goto.google.com/bisectoutput
Also note that this is blocked on the cleanup of the nested steps feature.

 Issue 396354  has been merged into this issue.

Closing out old bugs, re-open if still valid.