Confused by some numbers from gradle-profiler. I'm trying to benchmark what a typical "rebuild" of an android app would be after I add a line

println()

and rebuild. If I do this like 10 times in a row manually I get about a ~4 second build time. If I use gradle profiler with

apply-non-abi-change-to

("Change the body of a public method in a Java or Kotlin source class.") the lowest time I'll see is a 7 second build time over the 10 default runs (with default 6 warm ups). Anyone have a clue on how to debug why it seems like like build time is consistently close to twice of what "real world" testing shows?

have you tried to compare build scans?

i have not compared build scans. it should use my gradle properties in both situations so it should be using the daemon. i agree that 4 to 7 isn't huge, but it is in this case, and im a bit more worried why i can't repro a 7 second build in AS, meanwhile it seems to repro easily in gralde-profiler.

First thing you should really compare build scans and check the differences

interesting. i wonder how i can run a build scan on a gradle-profiler but i will do that. thank you. you're probably right that theres some sort of difference.

> In order to create a build scan of your build, use

--profile buildscan

. The build scan URL is reported on the console and is also available in

profile-out/profile.log

. https://github.com/gradle/gradle-profiler?tab=readme-ov-file#gradle-build-scans

thanks i will try to double check all of my settings here. appreciate it andrey