Performance regression due to blocking out constant propagation

[arikheinss]

Hello once more,

I noticed another source of performance regression when using partial functions compared to native anonymous functions x -> ...., which is that partial functions may lock the compiler out of some of its optimization routines. I documented my findings in details in a topic on discourse here.

As mentioned in the topic, AFAIK there is no solution to this problem that could be employed for your package, but I thought as the maintainer maybe you would want to know about the issue.

That is all for now. Cheers, and merry christmas!

[archermarx]

Thanks for this. Hadn't seen it until now and you definitely put a lot of thought into it. Do you know if the state of things has changed in the intervening year and a half?

[arikheinss]

Hi. Sorry for the late reply from me too.

AFAIK nothing in this regard has changed, but I'm no longer very up to date in the Julia ecosystem. I have switched Jobs since then and we do not use any Julia (sadly). If I do, I typically default to the package Underscores.jl, where this isn't really a question.