fix: Stacktrace is lost with one assertion in AssertSoftly, #4171 #4432
Conversation
...s/kotest-assertions-core/src/jvmTest/kotlin/com/sksamuel/kotest/matchers/AssertSoftlyTest.kt
Outdated
Show resolved
Hide resolved
| message.shouldContainInOrder( | ||
| "The following assertion failed:", | ||
| "1) Expecting actual not to be null", | ||
| " at com.sksamuel.kotest.matchers.AssertSoftlyTest${'$'}1${'$'}1${'$'}9.invokeSuspend(AssertSoftlyTest.kt:162)" |
There was a problem hiding this comment.
Minor but blocking: let's remove line number 162 from expected string number 3, to make the test less fragile. Otherwise any refactoring/reformatting will break this test.
There was a problem hiding this comment.
That makes sense. However, this was one of the problems reported in the issue: an incorrect pointer in the stack trace. I'll move that test to a separate class to prevent it from affecting the other tests. What do you think?
There was a problem hiding this comment.
That makes sense. However, this was one of the problems reported in the issue: an incorrect pointer in the stack trace. I'll move that test to a separate class to prevent it from affecting the other tests. What do you think?
I concur - let's extract this test into its own file.
There was a problem hiding this comment.
Please take a look now, I changed the hardcoded line number to dynamically calculated variable.
There was a problem hiding this comment.
Please take a look now, I changed the hardcoded line number to dynamically calculated variable.
LGTM, started the build
There was a problem hiding this comment.
Please take a look now, I changed the hardcoded line number to dynamically calculated variable.
the following test is failing, and it might be related: "adds an Exception to an empty collection of assertion failures"
There was a problem hiding this comment.
Yes, I see that there is a test class marked to run on Linux only, which is why I didn't catch it failing in my local run.
I've already fixed it. The issue was that the expected exception class had changed, since any assertion failure in the assertSoftly block should now be classified as a MultiAssertionError, even if only one assertion fails.
| it.message.shouldContainInOrder( | ||
| "The following assertion failed:", | ||
| "1) Expected map to match all assertions. Missing keys were=[], Mismatched values were=[(key, \"hi\" should have length 4, but instead was 2)], Unexpected keys were [].", | ||
| " at com.sksamuel.kotest.matchers.maps.MapMatchersTestKt${'$'}matchMapTests${'$'}1${'$'}1${'$'}3.invokeSuspend(MapMatchersTest.kt:596)", |
There was a problem hiding this comment.
Same suggestion: can you remove 596 to make the test less fragile?
pientaa
force-pushed
the
fix/4171-stack-trace-lost-assert-softly
branch
from
345b6c1
to
d943e3c
Compare
It looks like
MultiAssertionErrorfactory method already handles properly following cases:size == 1 && subject != nullsize == 1 && subject == nullSo the easiest fix is to use
MultiAssertionErroreven though there is just one assertion error - this way we don't loose the stacktrace. I provided appropriate test which covers the case presented in issue #4171