Migrate pp system implementation to osuAkatsuki/akatsuki-pp-rs

[7mochi]

Closes #283, closes #184, closes #101

[minisbett]

Why would you make the AR factor 0 on relax

[minisbett]

This is the worst relax implementation I've ever seen can't we just implement the Akatsuki aim crosscheck instead

[minisbett]

Nvm it's the bancho implementation I forgot it sucks

[minisbett]

We will definitely have to fork rosupp i don't think anyone here wants 1k pp plays getting nerfed to 200-300pp

[tsunyoku]

tbh just use https://github.com/osuAkatsuki/akatsuki-pp-rs since it's rosu-pp too.
just note the relax system lies in osu_2019 and not osu, because adapting 2022 for relax was awful

i'd also like to note that rosu_pp_py is not up to date with the latest rosu-pp version (due to rewrite stuff), so i don't think making this change is a good idea right now

[minisbett]

Also better for customization so for this PR I would suggest putting in rosupp-py as a submodule instead

[minisbett]

tbh just use https://github.com/osuAkatsuki/akatsuki-pp-rs since it's rosu-pp too. just note the relax system lies in osu_2019 and not osu, because adapting 2022 for relax was awful

i'd also like to note that rosu_pp_py is not up to date with the latest rosu-pp version (due to rewrite stuff), so i don't think making this change is a good idea right now

Or that would work

[7ez]

just replace vn pp with rosu-pp, I don't see a sense in replacing the whole std calc with it

[tsunyoku]

just replace vn pp with rosu-pp, I don't see a sense in replacing the whole std calc with it

because we get to use a single system across the board which makes life much easier
also oppai is problematic & peace is outdated and doesn't support converts

however, we need to wait for rosu-pp-py to be updated at the very least; unless we decide to just use akatsuki's version before then

[7mochi]

tbh just use https://github.com/osuAkatsuki/akatsuki-pp-rs since it's rosu-pp too. just note the relax system lies in osu_2019 and not osu, because adapting 2022 for relax was awful

i'd also like to note that rosu_pp_py is not up to date with the latest rosu-pp version (due to rewrite stuff), so i don't think making this change is a good idea right now

I could try adapting that version with the latest rosu-pp changes. Would that work?

[tsunyoku]

tbh just use https://github.com/osuAkatsuki/akatsuki-pp-rs since it's rosu-pp too. just note the relax system lies in osu_2019 and not osu, because adapting 2022 for relax was awful
i'd also like to note that rosu_pp_py is not up to date with the latest rosu-pp version (due to rewrite stuff), so i don't think making this change is a good idea right now

I could try adapting that version with the latest rosu-pp changes. Would that work?

it already includes them, it’s up to date with rosu-pp’s main branch

[minisbett]

Either way, I still suggest putting rosupp-py in as a submodule so it can be modified

[cmyui]

nice work so far, thanks for taking this responsibility on!

[cmyui]

also, looks like i must've broke pre-commit ci at some point 🤔

[7mochi]

Which python bindings would I use with akatsuki-pp-rs?

however, we need to wait for rosu-pp-py to be updated at the very least;

I was using rosu-pp-py in my fork and it was working without problems D:

[cmyui]

Which python bindings would I use with akatsuki-pp-rs?

@tsunyoku if you could help him out a bit here would be very epic 11/10

[7mochi]

Which python bindings would I use with akatsuki-pp-rs?

@tsunyoku if you could help him out a bit here would be very epic 11/10

akatsuki-pp-rs doesn't have python bindings yet so idk know how I should proceed 💀
I guess I'll wait until there is one because I haven't been able to adapt those changes to rosu-pp-py on my own

[minisbett]

Which python bindings would I use with akatsuki-pp-rs?

@tsunyoku if you could help him out a bit here would be very epic 11/10

akatsuki-pp-rs doesn't have python bindings yet so idk know how I should proceed 💀 I guess I'll wait until there is one because I haven't been able to adapt those changes to rosu-pp-py on my own

Since akatsuki-pp-rs is based on rosu-pp shouldn't the normal rosu-pp bindings work? Idk how python bindings work but if I imagine it correct it's basically just saying access these and these methods and the methods should have the same signatures

[7mochi]

I think I got it to work, but there's no bindings repo in osuAkatsuki so i can't make a PR yet, anyways, i've tested it in my fork and it works. Could you guys check it? @cmyui @tsunyoku

Changes:

• rosu-pp-py: osuperu/rosu-pp-py_old@7032971
• akatsuki-pp-rs: osuperu/rosu-pp_old@cd002b0
• bancho.py (performance.py | calculate_performances_std)

if score["mods"] and Mods.RELAX & score["mods"]:
    calculator.set_acc_2019(acc)
    result = calculator.performance_2019(map)
else:
    result = calculator.performance(map)

[cmyui]

nice, @tsunyoku is working on making bindings for akatsuki's pp system

[cmyui]

i'll review this shortly

[cmyui]

@7mochi hey, do you have any uncommitted changes? i'd like to take over the pr for a bit when you're ready

[cmyui]

going to refactor the calculate_performances_x functions into a singular interface, now that we can

[cmyui]

I think we're good to go on this.

Can I get a re-review @tsunyoku, @7mochi please!

[cmyui]

also @alowave223 you should take a look, i think

[cmyui]

ah, found some bugs with callers of calculate_performances still using params that were moved into the scores param

[cmyui]

ok, things look good now. feel free to review away!