does the latest minutest version change how tests ...

# minutest

[christophsturm]

does the latest minutest version change how tests are named? I have this in my pitest config targetTests.set(setOf("randolf.*")) and if I bump the minutest config it does not find tests anymore

[christophsturm]

with 1.9.0 it works

[dmcg]

What is pitest config @christophsturm?

[christophsturm]

oh. pitest is a mutation testing framework

[christophsturm]

https://github.com/christophsturm/randolf/blob/master/build.gradle.kts#L108

[christophsturm]

and i thought maybe minutest 1.10 changed how test names are reported, so the test name filtering in pitests no longer works as expected

[dmcg]

Ah yes. It wouldn’t shock me if I had broken that, although I don’t remember having to change any of the tests that do check naming since 1.9

[christophsturm]

pitest is a great tool btw.

[dmcg]

I’ve heard good things, but didn’t know it could be applied to Kotlin.

[dmcg]

Please let me know if I can help with the test naming.

[christophsturm]

if you want to see the problem in action you can check out randolf master, change the minutest version to 1.10 and run ./gradlew pitest

[christophsturm]

https://github.com/pitest/pitest-junit5-plugin/blob/master/src/main/java/org/pitest/junit5/JUnit5TestUnitFinder.java thats probably where the tests are discovered

[dmcg]

Ah, that kind of makes sense. Previously I was relying on the default behaviour for the test identifier - the result of which was (I think) that there was a MethodIdentifier pointing at minutests() in JUnit5Minutests - the @TestFactory.

[dmcg]

Now I’m explicitly setting a file url to point to the source where the test and context methods are invoked. This lets IntelliJ jump to the actual source - but I bet will result in the TestIdentifier not having a MethodSource.

[christophsturm]

can you just set the methodsource too?

[christophsturm]

or maybe trigger the old behavior with a property? pitest only runs from gradle

[christophsturm]

i really like my test setup, but it has a lot of moving parts, minutest, pitest, pitest-gradle, pitest-junit5-plugin

[dmcg]

I could, but it would be better to make pitest better? What does it do with the information?

[christophsturm]

hmm i can file an issue in the pitest-junit5 repo

[christophsturm]

i think it uses the method source for the test name filtering

[christophsturm]

isnt it the right thing to do to set the MethodSource?

[dmcg]

There isn’t a method, just the invocation of ‘test’

[christophsturm]

i think this is the check that fails:

.filter(testIdentifier -> testIdentifier.getSource().isPresent())

[christophsturm]

and i think i can fix it in the pitest junit plugin

[christophsturm]

submitted an issue in the junit plugin https://github.com/pitest/pitest-junit5-plugin/issues/39

[christophsturm]

once thats fixed i can add pitest to the minutest build if you want

[christophsturm]

I looked into it and junit5 is a really confusing api. maybe the easiest way is to set the method source again?

[dmcg]

You’re not wrong about the API

[dmcg]

I don’t think that I can have a MethodSource URI and a file one.

[christophsturm]

ahhh

[christophsturm]

ok

[christophsturm]

i think i have a fix for the junit plugin

[dmcg]

👍

[christophsturm]

but i have no idea how to use my locally built jar file with gradle

[christophsturm]

do i need to add a mavenLocal() source?

[dmcg]

I think you can just do ./gradlew install

[dmcg]

and it will install it into ~/.m2

[christophsturm]

its a maven project but I’ll try

[christophsturm]

ok it works locally for me and i submitted a PR

[christophsturm]

hmm it seems it doesnt really fix it. maybe we need a separate pitest plugin for minutest.

[christophsturm]

minutest-junit5 is not really a lot of code

[christophsturm]

the test identifiers seem to be exactly the same with 1.10 and 1.9 so it must be something different.

Untitled

[dmcg]

TBH I’m a bit surprised that PiTest worked at all with Minutest - I would have expected it to be quite sensitive to the form of test specification.

[dmcg]

Another thought - I upgraded the version of JUnit that Minutest uses between 1.9 and 1.10. (5.4.0 -> 5.5.2). That might mean that you are getting a later version with some other incompatibilty?

[christophsturm]

in the randolf project i have every dependency bumped to latest (also junit) and it works

[dmcg]

So that’s Minutest 1.9 running against JUnit 5.5.2? Hmm, back to the drawing board

[christophsturm]

I’m suprised that it shows method source for minutest when you said you changed it.

[christophsturm]

do you think its easy to serialize the output of the junit api to json and back? then i could just use that for a pitest-junit unittest

[dmcg]

I’m afraid I really don’t know what pitest is doing with the information, and I’m a bit snowed under. Are you going to be at KotlinConf - we could look at it together there?

[christophsturm]

sadly not. I think I’m just going to create a pitest minutest plugin then i can use tdd to do it instead of manually trying out things, which i don’t really like

[christophsturm]

kotlintest has a pitest plugin and its also apache license so i could use that as a starting point

[christophsturm]

probably less fun than starting from scratch, but a pragmatic solution