When running

app:build

task the

detekt

task runs as well. How can I disable this behaviour?

build

runs

check

which runs

detekt

. It is usual that build runs almost everything, what are you trying to do? maybe you want to compile + test, so

./gradlew assemble test

would be enough

Yea I agree with Javier, looking at the picture here will give you more clear view: https://www.google.com/url?sa=i&url=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F52[…]e=images&cd=vfe&ved=0CBAQjRxqFwoTCNDBhtCnkv4CFQAAAAAdAAAAABAD

At first glance it looks like running tests and linters with Gradle

check

(defined as

check

task dependency) task is great idea. However this has one major downside - when PR checks build fail you can't instantly determine the reason of the failure - you have to you have to manually read through (usually very long) log file. The solution for this is to runn all check in parallel (see screenshot). Other linters such as

spotless

are solving this issue by introducing configuration parameter:

spotless {
    isEnforceCheck = false
}

Instead of heaving dedicated parameter Detekt proposes disabling this check by filtering tasks:

tasks.named("check").configure {
    this.setDependsOn(this.dependsOn.filterNot {
        it is TaskProvider<*> && it.name == "detekt"
    })
}

I am using Gradle conventions plugins to setup detekt and for me the above solution does not work

1. you can use

--continue

so that Gradle runs as many tasks as it can even if some fail (only dependent tasks will be cancelled). the final build will still fail if there were any failures. 2. detekt emits reports so you can just grab those. and/or enable SARIF output, it shows up directly in the GitHub UI, https://detekt.dev/docs/introduction/reporting/#integration-with-github-code-scanning

Continue does not solve the problem (detekt has

ignoreFailures

option). The main issue here is lack of visibility - you have to dig deep to understand which check has failed and why (detekt in not the only verification task, so there is very long build log and bunch of reports).

Why running build filtering tasks instead of running each of those task individually?

I run different tasks to verify different aspects of the project separately: • build - does the project builds • detekt - check detekt • tests - run tests

I had a similar setup and I was running in parallel each individual check + assemble

BTW I am currently can run detekt with

build

but this is redundant as build task fails because of detekt issue - I would like to have more separation

I had this setup: • assemble (compile sources) • test • detekt • spotless • qodana

in Gradle (inherited from Maven terminology), "build" intentionally includes tests. "assemble" is "build but not test".

yeah, the name build is misleading

but I think

--continue

helps reduce duplicated work. instead of having multiple jobs that all build the same code, just have one job that does as much as possible. Gradle can still work in parallel (although you need to enable configuration cache to get the full benefits)

still running in parallel can help to finish the checks earlier

I agree this is always a tradeoff. I have private dedicated server, so this extra visibility benefit adds minimal cost for me. BTW @Javier comment about misleading name is actually very good one, I will investigate this

build basically runs almost everything as it is assemble + check, so it compiles the code, run the tests, run all verification tasks (spotless, detekt, etc). There are some exceptions, for example Android integration tests don’t run with build and must be run manually.

FYI Detekt is considering to add a new config option

detekt {
    isEnforceCheck = false
}

https://github.com/detekt/detekt/issues/5965#issuecomment-1497392847