<https://youtrack.jetbrains.com/issue/KT-45704>

# jvm-ir-backend-feedback

[eygraber]

https://youtrack.jetbrains.com/issue/KT-45704

[louiscad]

Did you test on 1.5-M2 and RC? I see the affected version only mentions 1.3.72 and 1.5-M1.

[eygraber]

I tested on M2 and RC but didn't get a chance to push the topic.

[louiscad]

"push the topic"?

[eygraber]

Update it. I figured that since it was being tracked whoever was working on it would know that it wasn't fixed yet.

[louiscad]

So, same issue on these versions? I think it'd be helpful to click on "affected versions" and add these.
BTW, I'm a bit surprised: happening on 1,3,72 and 1.5… what about 1.4.x?

[eygraber]

Oh interesting, didn't know I could change those. I tested on 1.4.x and 1.5 M1 but not on 1.3.72. Not sure how that got there.

[louiscad]

You can also edit an already submitted issue as well if you're the author. That can be used to add details that help the people dealing with your issue for example.

[udalov]

Just wanted to clarify that our highest priority right now are issues which started to reproduce with JVM IR in 1.5.0, since the effect is that a project working with 1.4.32 will fail to compile if the user updates Kotlin to 1.5.0.

Since the issue as stated in KT-45704 has been there apparently since 1.0.0, it’s not the highest priority, but we’ll get to it later. Please correct me if I misunderstood something.

It makes me wonder though how does your project, if it indeed contains such code pattern, compile right now? Is there any chance you’ve reported another problem that was found accidentally during minimizing the original problem?

[eygraber]

This code works fine in 1.4.32 unless we turn on IR, and it is also broken on 1.5

[udalov]

Sorry if I'm being unclear. The code as provided in KT-45704 fails in 1.4.32 too (with the same IOOBE "Cannot pop operand off an empty stack"):
https://pl.kotl.in/jXXL81YOi
... and with minor modifications it fails on all earlier Kotlin versions even including 1.0.0.

So my suspicion is that while minimizing the issue, you've accidentally changed the code so that it no longer represents the actual problem you have in your project, but fails with this irrelevant error that has been there forever.

[eygraber]

Hmm it's possible that happened. I'll investigate. If that is the case, can I DM you the original code?