Also, during the call with @ dmitriy.novozhilov, he mentioned that we could make our own DCE that will be based only on used symbols (classes/functions/fields). AFAIU, DCE in K/JS is much more powerful and it can eliminate dead code even inside function bodies, so maybe we don't need such complexity and symbols-level DCE will be quite enough for us. It can speed up our tests since Python will take less time to parse cut file

Considering the box tesrs performance, the time that we save on parsing long file by Python may go to DCE-ing itself 😅 but yeah, finally it would be good to have some DCE

the time that we save on parsing long file by Python may go to DCE-ing itself

I would say it's arguable: both parsing and DCE-ing are complex tasks and for me it's not clear which is more performant. Anyway, currently our tests generate DCE version of output too, so we just execute both parsing of a big file and complex DCE-ing 😅

Do they generate both ? I didn't know :D

Just take a look at the dirs in the generated testData dir. There are both full and dce-ed versions, and also there is one more (don't remember which one but probably a "minified" version)

Thanks, will check it out 👍

I've disabled dce and pir and it reduces tests time from 9 mins to 7 mins for me. Let's see what will be on CI: https://github.com/krzema12/kotlin-python/commit/23ffa1580ad579623e4ad707f90fd2b69be6b19e