When running benchmarks in the lab, it is ideal to fail fast and avoid using resources unnecessarily. When a benchmark fails we would like to acknowledge the developers as soon as possible.
Currently, the harness keeps executing all benchmarks to completion. Is there a way to tell BDN to fail fast?
I think that it would be nice to add such feature.
It could be added to IConfig as StopOnFirstError with default value = false and exposed via console line arguments to override.
@wojtpl2 maybe you would like to give it a try?
Ok next week
On Wed, Oct 31, 2018, 19:13 Adam Sitnik <[email protected] wrote:
I think that it would be nice to add such feature.
It could be added to IConfig as StopOnFirstError with default value =
false and exposed via console line arguments to override.@wojtpl2 https://github.com/wojtpl2 maybe you would like to give it a
try?—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/dotnet/BenchmarkDotNet/issues/939#issuecomment-434788204,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AQh-j4rdJNIUyt1voRgaBfZ5LH7MbUK7ks5uqehigaJpZM4YEowc
.
@adamsitnik What is the reasoning for default value = false?
@jorive to don't change the current default behavior (breaking change). I like the current one: last night I started the benchmark run for all the benchmarks we have and today in the morning I had the results for few thousands of them. 8 failed had no results but I had the errors in the logs. If it would stop on the first error I would have to fix the problem and rerun all the benchmarks again. With current behavior, I just fixed and re-run 8 of them.
Fixed by #947
Most helpful comment
@jorive to don't change the current default behavior (breaking change). I like the current one: last night I started the benchmark run for all the benchmarks we have and today in the morning I had the results for few thousands of them. 8 failed had no results but I had the errors in the logs. If it would stop on the first error I would have to fix the problem and rerun all the benchmarks again. With current behavior, I just fixed and re-run 8 of them.