I'm running a scenario like this:
build-warm {
title = "build-warm"
tasks = ["app:buildInternalDebug"]
warm-ups = 4
iterations = 5
}
After upgrading the profiler version from 0.23.0 to 0.24.0, the task start metric seems to accumulate the total execution time (or something along these lines) and isn't reset between runs. The total execution time itself is not affected though.
Example results:
0.23.0 (median task start is 0.792 s, as expected)

benchmark-0.23.0.csv
benchmark-0.23.0.html
0.24.0 (median task start is 36.383 s)

benchmark-0.24.0.csv
benchmark-0.24.0.html
The only difference between these runs is the version of the profiler.
This seems to impact only scenarios that don't do any mutations. For example, a scenario like this is not impacted:
build-non-abi-change-incremental {
title = "build-non-abi-change-incremental"
tasks = ["app:buildInternalDebug"]
apply-non-abi-change-to = "path/to/file.kt"
warm-ups = 4
iterations = 5
}
I'm running a scenario like this:
After upgrading the profiler version from 0.23.0 to 0.24.0, the
task startmetric seems to accumulate the total execution time (or something along these lines) and isn't reset between runs. Thetotal execution timeitself is not affected though.Example results:

0.23.0 (median
task startis 0.792 s, as expected)benchmark-0.23.0.csv
benchmark-0.23.0.html
0.24.0 (median

task startis 36.383 s)benchmark-0.24.0.csv
benchmark-0.24.0.html
The only difference between these runs is the version of the profiler.
This seems to impact only scenarios that don't do any mutations. For example, a scenario like this is not impacted: