osu-lazer

Ranked Play: Fix stutter when entering queue screen (#37295 )

The panels look up the online APIUser models. Maybe I could do this by
doing the lookups async inside `RankedPlayMatchPanel`, but this will
probably do for now?

Dan Balasescu · 2026-04-15 01:10:07 +09:00

daff79b1d3

Ranked Play: Fix chat temporarily appearing during intro (#37294 )

I was kind of lazy with the disappear/appear stuff. Made it properly set
the required state at the correct time now.

Made a second fix to change it into a visibility container, so that
bindable states are deduped. On `master` it would re-appear from the
bottom with every stage change.

Dan Balasescu · 2026-04-15 00:28:26 +09:00

ca40749a14

Add bottom 'ornament' overlay to ranked play (#37288 )

This commit adds an always present overlay to all ranked play screens,
meant to indicate to the user that a ranked play session in currently in
progress.

This has been largely inspired by the pre-shader argon healthbar code.
It shouldn't have the same performance concerns, however, since the
paths are only calculated once when loading the drawable (and eventually
when it is resized, if ever).

`RankedPlayScreen`:

<img width="1838" height="1353" alt="image"
src="https://github.com/user-attachments/assets/c621d759-a88f-49f0-b0df-3a6da90eca65"
/>

Gameplay:

<img width="1838" height="1353" alt="image"
src="https://github.com/user-attachments/assets/ee848d06-3878-4cbb-bd4b-de2c4bcfa688"
/>

Currently the drawable overlaps with some components, but it will be
resolved in later pull requests.

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Krzysztof Gutkowski · 2026-04-15 00:27:43 +09:00

16a2a96bcf

Add additional information to the matchmaking queue screen (#37229 )

- [x] Depends on https://github.com/ppy/osu/pull/37226
- [x] Depends on https://github.com/ppy/osu-server-spectator/pull/464

This adds two new components to the queue screen:
- A listing of the most recently completed matches (global).
- A rank distribution graph.

It looks something like this (fake data / test scene):

<img width="1669" height="1005" alt="image"
src="https://github.com/user-attachments/assets/caa57119-4267-4c6e-9898-2f414de865bf"
/>

It's completely dev-design(TM), but I used Lichess as inspiration for
the graph, and the original design document as inspiration for the
panels.

As for the history, because these are _completed_ matches one of the
player life points will always be 0, but I've designed it so as to
possibly support showing ongoing matches too in the future. It's only
supported for ranked play right now, though there is no reason we
couldn't track quick play rooms too (it's just... I'm not sure how to
design the panels for quick play).

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Dan Balasescu · 2026-04-14 10:54:54 +09:00

ac620ee375

Fix multiple issues with ranked play card audio previews (#37262 )

The main goal here is to:
- Make keyboard selection play previews just like hovering does with
mouse.
- Make sure cards don't play previews when they are in animation.
- Drive by fix to fix toggling discard not working via keyboard.

(I really want to rewrite all these classes, the structure is not great)

---------

Co-authored-by: Dan Balasescu <smoogipoo@smgi.me>

Dean Herbert · 2026-04-12 15:43:47 +09:00

0e9664bfdf

Fix missing null check

Dan Balasescu · 2026-04-12 14:48:44 +09:00

bcc36b4884

Add more ranked play SFX (#37258 )

- Stage changes (with ducking):


https://github.com/user-attachments/assets/1ed1cf5a-759b-487a-9028-314accc15a6a

- Messages (with debouncing):


https://github.com/user-attachments/assets/c8c38d18-4fcc-4654-887b-be3de3e84949

Message sample might be a bit too clicky, but consider this a first
pass.

---
- [x] depends on ppy/osu-resources#417

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Jamie Taylor · 2026-04-12 01:43:36 +09:00

e36f9b2dcf

Fix lack of encode-decode stability when writing out mania beatmaps with some key counts (#37256 )

Closes https://github.com/ppy/osu/issues/37232.

The actual fix is
https://github.com/ppy/osu/commit/e959b20517497a093d3c00a17457c5d36bf57651;
everything else is window dressing / test harness to ensure I don't try
and do a wrong change like https://github.com/ppy/osu/pull/37251 did. I
recommend reviewing commit-by-commit.

See [this desmos](https://www.desmos.com/calculator/a5yjpacvxa) for
visual explanation of change, I think it does a better job at explaining
this than any words I could type here.

Of note:

- In the end this did only affect 14K but that should never be assumed
when floating point is involved.
- Test cases generated here were generated in stable manually.
- Except for 11 / 13 / 15 / 17K which are not officially supported and
which don't work in lazer due to orthogonal reasons (see comment added
in this PR in `ManiaBeatmapConverter`), decoding in lazer was always
fine.
- My worry was that the old encoding method before this PR could
potentially cause stable to move a note from one column to another but
thankfully that is not the case. The old method of encoding columns as X
positions does not cause issues wherein lazer reads them back
differently than stable after encode.

I checked this by checking out `master`, re-encoding all of the test
stair-pattern nK beatmaps added in this PR on `master`, exporting that
as compatibility, re-importing to stable, and cross-checking that the
decoded beatmap is visually the same on lazer and on stable.

This is important to check because if this wasn't the case, we'd
potentially have cases of actual online beatmaps (remember that we have
BSS now) wherein a beatmap plays differently on stable than on lazer due
to notes moving between columns, and would need to screen for this being
the case and potentially apply corrective / reconciliatory action.

Bartłomiej Dach · 2026-04-11 00:38:22 +09:00

6231e06ebe

Update editor icons (#37201 )

Brings in new fresh icons for the toolbox (select, hitcircle, slider,
spinner, grid) and toggles (new combo, whistle, finish, clap,
grid/distance snap).

Depends on https://github.com/ppy/osu-resources/pull/416.

| Before | After |
|--------|--------|
| <img width="300" height="897" alt="image"
src="https://github.com/user-attachments/assets/279ea162-c3e2-4312-a0a2-39ad08df16f9"
/> | <img width="304" height="889" alt="image"
src="https://github.com/user-attachments/assets/aab2a143-0e6b-45c1-bf84-8998c12bd960"
/> |

Krzysztof Gutkowski · 2026-04-10 20:27:21 +09:00

3e6233072c

Tidy up ResultsScreen (#37242 )

Just the bare minimum code quality so I can start working on these
classes..

Please push back if this doesn't seem better than what was already
there. This is mostly autopilot fixing for me based on how I've been
writing code for osu! to date.

There are changes to the load process but nothing which should cause
issues, I hope.

Dean Herbert · 2026-04-10 20:14:11 +09:00

4b8fa892ea

Improvements to discard screen UX (#37245 )

Just an initial grab bag to keep these PRs small.

### Avoid showing countdown update when at discard screen

This is needless. We already have the `DiscardFinish` stage which has a
short countdown. Playing this change to the user creates unnecessary
confusion.

### Allow stage caption text to be changed at any point

Also remove custom colour support. We'll handle this internally in a
better way in the future.

### Better explain why we're waiting after discarding our own cards

Dean Herbert · 2026-04-10 20:12:27 +09:00

9c43739228

Add matchmaking profile badge (#37241 )

Using the same styling as osu!web + daily challenge.

<img width="1920" height="1034" alt="Screenshot_20260409-164600"
src="https://github.com/user-attachments/assets/97e2270e-af9f-478d-b2d6-c9fb8be16720"
/>

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Dan Balasescu · 2026-04-09 19:25:21 +09:00

93b7c3324d

Make RankedPlayColourScheme properly static (#37244 )

Dean Herbert · 2026-04-09 18:22:32 +09:00

b838564039

Do not send replay frames to spectator server if initial begin play invocation failed (#37159 )

RFC

Until now, if the initial `BeginPlaySession()` call failed, the client
would continue operating as if it didn't - it would still continue to
send frames and call `EndPlaySession()` at the end of a session.

Server-side, two things generally can happen after this:

- The sent frames and the `EndPlaySession()` call are
[completely](https://github.com/ppy/osu-server-spectator/blob/7bab117e9d161455485368f63a0607a9e53f9f8a/osu.Server.Spectator/Hubs/Spectator/SpectatorHub.cs#L122-L125)
[ignored](https://github.com/ppy/osu-server-spectator/blob/7bab117e9d161455485368f63a0607a9e53f9f8a/osu.Server.Spectator/Hubs/Spectator/SpectatorHub.cs#L153-L157)
as no-ops, or
- A hub filter (like `ClientVersionChecker`) that failed the initial
`BeginPlaySession()` call continues to fail the calls to
`SendFrameData()` and `EndPlaySession()`, all the while creating a storm
in logs, because it needs to throw `HubException`s to communicate to
users that they need to update their game, and the exceptions can't be
silenced from logs because they look like every other failure.

To that end, this has two goals: reduce useless network traffic, and
reduce noise in spectator server logs after the client version checks
were recently reactivated.

Probably needs tests, but unsure if everyone's going to be on board with
this to begin with to be quite frank, so I'm leaving tests for when I'm
told this needs tests.

Bartłomiej Dach · 2026-04-09 03:40:32 +09:00

ca44d8233d

Add models for improvements to matchmaking lobby (#37226 )

1. Gives `MatchmakingJoinLobby` parameters.
2. Adds additional data to lobby status update models.

A further PR will build upon (2) to add more data to the queue screen.

Dan Balasescu · 2026-04-08 18:43:51 +09:00

2026.408.0 8c6818e275

Localise tabs on DashboardOverlay (#37236 )

Denis Titovets · 2026-04-08 14:46:14 +09:00

2058297a8d

Fix footer buttons still being able to fire their action when temporarily hidden by overlay content (#37231 )

In human words: I read [this forum
thread](https://osu.ppy.sh/community/forums/topics/2195138?n=1) today
and was horrified to see the user there opening the F3 options menu in
song select *when the F1 mod overlay was pulled up*, which is (a) not
intended UX, (b) looks terrible, and (c) just wrecks the game
behaviourally wholesale from start to end.

So with this change you don't get to open options via F3 while inside
mod overlay at all.

The `Action` shadowing is pretty ugly but I don't have better ideas.
Initially I tried to mess with `Enabled` (as I did once previously, see
https://github.com/ppy/osu/commit/36628e24f92c286b87c118c7c1bb9bc582895571),
but it's much more complicated in this case because the enabled state
needs to be restored when the buttons reappear, or it could change
independently while the buttons are temporarily hidden, etc. So I'd
rather just not deal with all that and invent a parallel scheme.

Bartłomiej Dach · 2026-04-08 00:40:50 +09:00

2026.408.0-tachyon 97ddc8e160

Tidy up track switching handling (#37230 )

This doesn't fix https://github.com/ppy/osu/issues/37136, but it does
fix the fact that I left some weird code around in
https://github.com/ppy/osu/commit/aab01b0fe54c543e015e7be4f218bf71d72290c7
that wasn't required (the re-equality check and disposal).

Just cleaning that up here. I doubt it will fix the issue. But also, I
can't see how the issue can occur yet.

Dean Herbert · 2026-04-08 00:03:05 +09:00

ed84dd431f

Improve efficiency of tag population (#37228 )

- No longer holds realm write transaction open while performing sqlite
lookups.
- No longer attempts a write transaction when it will be a noop.

I'll admit that this is maybe working around the actual realm write part
being slow, but as I can't profile the issue locally, the sluggishness
may actually be in sqlite for those users affected (since it's only been
reported for tag population and not difficulty calculation?).

Regardless, this should fix the issue this iteration.

I also adjusted the user messaging to let them know why tag population
is happening, since we've had some questions as to why it's running in
the first place (it only happens once a month, so that's
understandable).

- [x] Depends on https://github.com/ppy/osu/pull/37227.
- Closes https://github.com/ppy/osu/issues/34699.
- Closes https://github.com/ppy/osu/issues/37210.

Note that https://github.com/ppy/osu/pull/36128 also exists and has
valid improvements which can be addressed separately. This is intended
to be something we can act on immediately.

---------

Co-authored-by: Bartłomiej Dach <dach.bartlomiej@gmail.com>

Dean Herbert · 2026-04-08 00:02:32 +09:00

84cce2f0d1

Add ability to search for users (#37225 )

This was a private request for roundtable event usage. It’s also a
common feature request, so I decided to spend a bit of time getting this
working well-enough.


https://github.com/user-attachments/assets/acceb57f-2979-43d0-9fc2-33e977bd2dd5


---

### Delay loading spinner / loading layer initial load briefly to avoid
flickering

There's cases in this overlay where loading takes a few milliseconds.
The loading spinner gets annoying. This also happens elsewhere, so this
could be considered a global fix. Separate PR? probably...

### Ingest loading state of dashboard child content to show more correct
loading layer

Each display had their own loading layer implementation, but this is
already too deep (inside the scroll content) and doesn't display great
when for instance, results don't take up the full screen height.

---------

Co-authored-by: Bartłomiej Dach <dach.bartlomiej@gmail.com>

Dean Herbert · 2026-04-07 14:14:49 +02:00

c570db6c40

Add method for silently dismissing progress notifications without completion (#37227 )

This was used in one place, but I foresee this being a more common
scenario. This also fixes an edge case where the dismiss process would
fail if completion happened on an async thread before the
`NotificationOverlay`'s scheduler could handle the initial ingress.

Dean Herbert · 2026-04-07 13:43:14 +02:00

a3a530c25d

Ranked play general UI/UX fixes (#37220 )

Dean Herbert · 2026-04-07 15:33:29 +09:00

e80e5f37ec

Ranked Play: fix cards not going all the way off-screen (#37222 )

`RankedPlaySubScreen.CenterColumn` had a padding which moved the card
hand up slightly, causing it to not fully dissapear when contracting.

The padding doesn't serve any purpose anymore (remnant of the very early
versions of the screens), so I just removed it.
I checked against `DiscardScreen`, `PickScreen`, `OpponentPickScreen` &
`EndedScreen` to make sure this doesn't cause any layout breakage.

Also removed the `ButtonsContainer` since it isn't being used anywhere
anymore.


https://github.com/user-attachments/assets/2fd32407-fbf7-45a3-b92a-0730a0f8a3fd

maarvin · 2026-04-06 23:05:42 +09:00

ea0c04ae4f

Fix skin editor origin dropdown options applying origin in wrong coordinate space (#37219 )

Closes https://github.com/ppy/osu/issues/37185.

The checkboxes in the context menu's ternary states were supposed to
always show the origin "in local space", even if anchor is set to
"closest". The issue here was reusing a method that only really made
sense with closest anchor active for explicit application of
"local-space" origin.

To recap:
- Below I will use concepts of "local-space origin" and "screen-space
origin". To understand the difference, let's use an example:

Say there's a drawable with 180 degree rotation. Suppose it has the
"local origin" of `TopCentre`. The "local origin" is just the `Drawable`
notion of origin; you'd literally set `d.Origin = Anchor.TopCentre`.

The "screen-space origin" of this drawable is `BottomCentre`, because
due to the rotation, that's how the component will visually behave when
its position is altered.

The same sort of distinction applies forth to things like flips /
negative scale and such.

- When you have closest anchor selected, you can only choose the anchor
to snap to. The drawable will snap to that anchor, and choose an origin
closest to it *in screen space* such that the "closest" in "closest
anchor" works as users would expect it to. In this state, if you open
the context menu for origin, all items will be disabled, but the ternary
menu items will show the origin state *as translated back to local
space*.

- When you have an explicit anchor selected, you can choose both the
anchor and origin. In that case, the origin picked is always picked in
local space.

In the end, this is all consistent with how the `Origin` property on
`Drawable` works, and also with what is serialised to skin jsons.

Bartłomiej Dach · 2026-04-06 23:03:07 +09:00

0f40d61bb8

Turn tests green, for a short while, maybe (#37218 )

## [Rewrite `BackgroundMusicManager` to not run into framework
breakage](https://github.com/ppy/osu/commit/622216d8911832c39fa4e126b2810e4e0f46cbf7)

The attempted proper fix to this was
https://github.com/ppy/osu-framework/pull/6727. Unfortunately when
presented with [the framework
bump](https://github.com/ppy/osu/pull/37217) with that change, CI says
"you're stupid" and fails on some disposal idiocy that of course is
undebuggable and irreproducible:

The active test run was aborted. Reason: Test host process crashed :
Unhandled exception. System.AggregateException: One or more errors
occurred. (Object reference not set to an instance of an object.)
---> System.NullReferenceException: Object reference not set to an
instance of an object.
at osu.Framework.Audio.Sample.SampleChannelBass.Dispose(Boolean
disposing)
at System.Threading.ExecutionContext.RunInternal(ExecutionContext
executionContext, ContextCallback callback, Object state)
	--- End of stack trace from previous location ---
at System.Threading.ExecutionContext.RunInternal(ExecutionContext
executionContext, ContextCallback callback, Object state)
at System.Threading.Tasks.Task.ExecuteWithThreadLocal(Task&
currentTaskSlot, Thread threadPoolThread)
	   --- End of inner exception stack trace ---
	   at osu.Framework.Audio.AudioCollectionManager`1.UpdateChildren()
	   at osu.Framework.Audio.AudioCollectionManager`1.UpdateChildren()
	   at osu.Framework.Audio.AudioCollectionManager`1.UpdateChildren()
	   at osu.Framework.Audio.AudioCollectionManager`1.UpdateChildren()
	   at osu.Framework.Threading.AudioThread.OnExit()
at osu.Framework.Threading.GameThread.setExitState(GameThreadState
exitState)
	   at osu.Framework.Threading.GameThread.RunSingleFrame()
at osu.Framework.Threading.GameThread.<createThread>g__runWork|70_0()
at System.Threading.ExecutionContext.RunInternal(ExecutionContext
executionContext, ContextCallback callback, Object state)
	--- End of stack trace from previous location ---
at System.Threading.ExecutionContext.RunInternal(ExecutionContext
executionContext, ContextCallback callback, Object state)


(https://github.com/ppy/osu/actions/runs/24019928154/job/70046733058?pr=37217#step:5:119)

I no longer have the energy for any of this shit.

@nekodex would appreciate if you could check that I actually haven't
broken anything with the bgm here. Seems okay to me in test scenes at
least.

## [Apply lowest-effort maybe-fixing changes to a bunch of flaking
tests](https://github.com/ppy/osu/commit/7bd3ca4adfcce5b90add11565a13f3fe9177ad5e)

None of the failures are reproducible locally, of course. I'm tired of
this. If anyone else wants to subject themselves to actually
investigating any of these, by all means, godspeed and good luck.

Bartłomiej Dach · 2026-04-06 17:59:39 +09:00

2026.406.0 809298ddeb

Ignore one more false positive rider issue

Dean Herbert · 2026-04-06 14:23:36 +09:00

2026.406.0-lazer db92681e51

Remove unused lookup component

Dean Herbert · 2026-04-06 14:23:23 +09:00

dfeb9d8321

Add current stage overlay to ranked play (#37202 )

Displays the stage name and details like currently picking player
and damage multiplayer where applicable.

Currently only shown on the discard and pick stages.

# Move user retrieval to `RankedPlayScreen`

This is done because I need the relevant `APIUser` instances in order to
pass them to the overlay component. `RankedPlayScreen` seems like the
appropriate place to manage the overlays since it manages the stage
subscreens, hence the need to access the `APIUser`s in here.

# Add current stage overlay to ranked play

The actual change of this PR. Very much a dev design.


https://github.com/user-attachments/assets/2388e934-2fc7-4e15-9947-9f98412765d2

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Krzysztof Gutkowski · 2026-04-06 03:00:58 +09:00

9a56aed1e9

Remove left over DI usage

Annoying my rider.

Dean Herbert · 2026-04-06 01:07:40 +09:00

b08a86f539

Display beatmap state in RankedPlayUserDisplay (#37188 )

This has gone through a few iterations, and eventually ended up as a
simple text percentage display next to the username. I feel that adding
another progress bar right next to the big healthbar would make things
too cluttered, and trying to move the beatmap state elsewhere would make
it too disconnected from the players that are potentially downloading a
beatmap.

I considered making the local user fetch download progress data using
`BeatmapDownloadTracker` instead of relying on `BeatmapAvailability` in
order to get more frequent updates, but that would add a lot of extra
complexity for little gain IMO.


[Screencast_20260403_095644.webm](https://github.com/user-attachments/assets/85fbd4b8-6b5c-41d2-b29b-c93885f73bb3)

Krzysztof Gutkowski · 2026-04-04 22:05:28 +09:00

8b69aa9fdb

Add SFX to ranked play results screen (#37193 )

It's a bit rough and some of the timing isn't perfect, but:


https://github.com/user-attachments/assets/bfee2fa5-b4cd-4665-b70f-d902ad93ae43

Also adds some silly placeholder sounds to the end screen:


https://github.com/user-attachments/assets/31e34067-4343-4464-9777-f93c8cad1021

---
- [x] depends on ppy/osu-resources#415

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Jamie Taylor · 2026-04-04 21:50:17 +09:00

ae955c0589

Add osu! (Debug, Second Client) launch option for VS Code (#37197 )

Kinda self explanatory, adds a second client configurations so its
easier to test multiplayer-specific things when using VSCode

For people that don't know how to use this, basically just run the first
debug like normal, then swap to the second client option and run that.
You can also do it in reverse. Visual guide here:

https://github.com/user-attachments/assets/1dab50eb-3bd2-422d-a776-852ac4454213

Tenexxt · 2026-04-04 11:16:55 +09:00

20573844f5

Assorted test deflaking (#37191 )

A few quick ones from https://github.com/ppy/osu/issues/37190.

## [Rewrite `TestSceneDeleteLocalScore` to have less context menu
containers (and hopefully no longer
flake)](https://github.com/ppy/osu/commit/ea0bc5b72374056d1749f8058f4874b7ef66a241)

As seen in
https://github.com/ppy/osu/actions/runs/23890777748#user-content-r3s3.

This is a speculative fix but I'm feeling somewhat confident about this
one.

`BeatmapLeaderboardWedge` has TWO separate `ContextMenuContainer`s
itself, and the test mentioned here was bringing a third. I have a
feeling that the test flaking may have something to do with the fact
that the test logic would attempt to click a menu item on specifically
ONE of the three context menus. My bet is that when it fails, it's
because it's trying the wrong one, but I don't have reproduction.

## [Wait for text to appear in flaking
`TestSceneDrawableRoomPlaylist.TestSelectableMouseHandling`
test](https://github.com/ppy/osu/commit/9d62eea8b56a0eb013e8116db0d38424b500876f)

As seen in
https://github.com/ppy/osu/actions/runs/23888446400#user-content-r1s1.

Speculative. Banking on the fact that it takes time to load the sprite
texts.

## [Remove all tests from `TestSceneWikiMarkdownContainer` dependent on
existence of wiki on
dev](https://github.com/ppy/osu/commit/8f319f91c98510b84e915bb61ec147618a584ce1)

Deletes a flaky as seen in
https://github.com/ppy/osu/actions/runs/23878899702#user-content-r2s2.

The year is 2026 and LLM scrapers hammer [the
entire](https://sourcehut.org/blog/2025-04-15-you-cannot-have-our-users-data/)
[internet](https://blog.metabrainz.org/2025/12/11/we-cant-have-nice-things-because-of-ai-scrapers/)
all over to scrape whatever ounce of Human Content there is left to feed
the Moloch so that it can regurgitate it back in the form of The Most
Average Speech You've Ever Read. We are not immune to this, and as such
the LLM homunculi have hit the dev.ppy.sh wiki instance enough times for
it to just completely [be banished to the
blagole](https://www.youtube.com/watch?v=AfA_2Ku1aJY).

Which means I get to freely delete flaky tests that should never have
been running as part of CI because they're completely useless now and
it's not like we're ever turning them back on again.

## [Use equality check in
`TestSceneBeatmapCarouselScrolling.TestScrollPositionMaintainedOnRemove_SecondSelected`
that's less sensitive to floating
point](https://github.com/ppy/osu/commit/6f2a1de58f7e816830c430e89117ee4c594f529d)

As seen in
https://github.com/ppy/osu/actions/runs/23826420794#user-content-r0s1.

Bartłomiej Dach · 2026-04-04 00:17:31 +09:00

090fe44f86

Fix ranked play test failures from adding null playlist items (#37189 )

Example:
https://github.com/ppy/osu/actions/runs/23900675414/job/69696255970?pr=37178#step:5:38

Regressed in https://github.com/ppy/osu/pull/37172, cc @LiquidPL

Would fail in multiple tests. I'm not going to spend time figuring out
exactly why, I'm just going to guess that not all tests bother to set up
the relevant playlist items for the cards or whatever.

Some of the failing tests are flaky but not because the `item` here
isn't sometimes null in those cases. It's always null, but the callbacks
are probably scheduled or whatever and therefore have a chance to never
run. Also some of the failures appear to cascade / spill from other
tests as well.

Bartłomiej Dach · 2026-04-03 22:11:50 +09:00

796877a594

Fix race condition in beatmap difficulty cache invalidation flow (#37178 )

Exposed by CI failures
([example](https://github.com/ppy/osu/actions/runs/23888446400#user-content-r0s0)).

The race occurs when a consumer calls `GetBindableDifficulty()` for the
first time and then a cache invalidation is triggered. The sequence of
events triggering the failure is as follows:

1. Consumer calls `GetBindableDifficulty()` to get a difficulty bindable
for a given beatmap tracking the game-global ruleset / mods. This
triggers difficulty calculation A.
2. In the meantime, another process requests a cache invalidation for
the same beatmap as the one supplied by the consumer in step (1). This
incurs a cache purge and triggers difficulty calculation B, but never
cancels difficulty calculation A.
3. Difficulty calculation B concludes and writes the correct, latest
difficulty value to the bindable.
4. Difficulty calculation A concludes and writes an incorrect, stale
difficulty value to the bindable.

See below for patch that reproduces this behaviour on my machine 100%
reliably:

```diff
diff --git a/osu.Game/Beatmaps/BeatmapDifficultyCache.cs b/osu.Game/Beatmaps/BeatmapDifficultyCache.cs
index https://github.com/ppy/osu/commit/d6b40639161e26af223f03761b3826b0cd7f4a87..c9604e0e58 100644
--- a/osu.Game/Beatmaps/BeatmapDifficultyCache.cs
+++ b/osu.Game/Beatmaps/BeatmapDifficultyCache.cs
@@ -252,17 +252,17 @@ private void updateBindable(BindableStarDifficulty bindable, IRulesetInfo? rules
             GetDifficultyAsync(bindable.BeatmapInfo, rulesetInfo, mods, cancellationToken, computationDelay)
                 .ContinueWith(task =>
                 {
+                    StarDifficulty? starDifficulty = task.GetResultSafely();
+
                     // We're on a threadpool thread, but we should exit back to the update thread so consumers can safely handle value-changed events.
-                    Schedule(() =>
+                    Scheduler.AddDelayed(() =>
                     {
                         if (cancellationToken.IsCancellationRequested)
                             return;

-                        StarDifficulty? starDifficulty = task.GetResultSafely();
-
                         if (starDifficulty != null)
                             bindable.Value = starDifficulty.Value;
-                    });
+                    }, starDifficulty?.Stars > 0 ? 400 : 0);
                 }, cancellationToken);
         }

```

The goal of the patch is to reorder the write to the bindable in order
to trigger the scenario described above.

Due to the invasiveness of the patch it is not suitable to add as a
test, and chances are that the schedule delay may need to be tweaked if
anyone else wants to check that patch.

Anyway, the solution here is to use the same pattern of creating a
linked cancellation token even on the first retrieval of a bindable
difficulty, and registering it in the list of cancellation tokens that
already existed to service the ruleset- / mod-tracking flow.

Some extra rearranging in
https://github.com/ppy/osu/commit/9184299239a8b5e82957d46db33f1d26bab238fd
is needed to ensure the linked tokens created to do this don't stay
behind after they are no longer needed for anything.

Bartłomiej Dach · 2026-04-03 21:57:07 +09:00

ce8406613a

Adjust multiplayer logic to accommodate for referees spectating in-game (#37152 )

- Last part of / closes
https://github.com/ppy/osu-server-spectator/issues/406.
- Remaining work on slots will be tracked in
https://github.com/ppy/osu-server-spectator/issues/405.

This PR is a corollary of
https://github.com/ppy/osu-server-spectator/pull/453 and all of the
dispensations referee users in a multiplayer have received therein. The
goal here is to allow access to all relevant room management functions
even if the referee in question isn't host, as well as to disallow
access to all non-relevant functions to do with the actual match
gameplay.

I'm not going to lie, this logic *is* ugly. I would argue that it
already *was* ugly on `master` and my goal was to operate with as light
a touch as possible myself. But you could see this as copping out and
that I should try to refactor some of this. I will try - but only after
someone else's seen the initial approach and deemed it unsuitable.

The logic in `MatchStartControl` is awful - there are so many moving
pieces of state that dictate what can happen when with all the buttons,
and yes, I am making it worse here.

This time there is some test coverage. Not everything is covered, but
the coverage should be on par in all components and pieces of relevant
logic I touched that already had tests covering them. On that note,
please forgive the diffstat size, but the tests *are* most of that size.

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Bartłomiej Dach · 2026-04-03 19:38:44 +09:00

31c1168d76

More tooling updates & fixes (#37177 )

## [Adjust CI test reporting to upstream action
changes](https://github.com/ppy/osu/commit/f736337b1acf311bf48fb8c39cae67c3f172608d)

It's been semi-broken since I bumped it a few weeks ago. Paper trail for
this is at https://github.com/dorny/test-reporter/issues/750, but to
TL;DR it: `dorny/test-reporter@>=v2.0.0` migrated from [creating new
check runs via the GitHub
API](https://docs.github.com/en/rest/checks/runs) to [job
summaries](https://docs.github.com/en/actions/reference/workflows-and-actions/workflow-commands#adding-a-job-summary)
by default.

The main difference is that compared to check runs, job summaries do not
*appear to* require extra `GITHUB_TOKEN` permissions, but in exchange
are limited to the job that called it. This broke visibility of the test
reports because due to `GITHUB_TOKEN` permissions foibles the test
reporter was running in a separate workflow.

I see migrating as a plus here, since:

- The visibility of results is comparable-to-better (example available
for preview at https://github.com/bdach/osu/actions/runs/23892493152)
- No longer required to have a completely separate workflow for test
result reporting
- No longer required to give `checks: write` permissions to the action
(I'd hope, we'll see, untested on a public repository with PRs involved)

One downside is there'll be no in-code annotations for failing tests
anymore but that's whatever IMO. Half the time they weren't even very
helpful, test results pretty much require maintainer interpretation
anyway.

This needs to be applied to a few other repos but I'm starting here
because this is the one where the traffic is highest and therefore
unbreaking the report is of most value (and also the one where I'll see
if it works with public PRs the fastest).

Side note, I was hoping to remove the artifact upload/download games by
just attaching the summary inside each individual test job in the
matrix, but [it looks like
crap](https://github.com/bdach/osu-framework/actions/runs/23888384309)
because only the first three summaries are loaded by default, so if
there are more, you have to click each remaining one to see its output.
Wow. Awesome.

Also updates the action to `v3.0.0` to resolve node deprecation
warnings.

## [Update inspectcode version to resolve deprecation
warnings](https://github.com/ppy/osu/commit/496cf6890ecb6571e6361731d46f53cceb7cb584)

More node deprecation warning fixes.

Bartłomiej Dach · 2026-04-03 00:19:46 +09:00

5b6d215583

Add "Synesthesia" mod for osu!catch ruleset (#37011 )

Mod that colours HitObjects based on the musical division they are on,
now in osu!catch


https://github.com/user-attachments/assets/2dda493d-dc8b-4ea4-ba47-7d04e2062b42

For now *bananas are not coloured by the mod and keep their yellow
colour*, since I think its better for the sake of readabilty (also it
just looks kinda ugly?). Do leave your thoughts on that.

Droplets are always coloured to `LightGreen`, setting their colour to
closest timeline ticks is wrong and looks incorrect since droplets
aren't generated with them in mind.

---------

Co-authored-by: Bartłomiej Dach <dach.bartlomiej@gmail.com>
Co-authored-by: Shavix <54279284+Shavixinio@users.noreply.github.com>

Tenexxt · 2026-04-02 23:22:46 +09:00

77ef450883

Fix osu!mania editor not zooming playfield to match timeline (#37176 )

- Fixes CI failures due to completely wrong test (sorry)
- Fixes #37167 (oops)

Dean Herbert · 2026-04-02 12:49:45 +02:00

e207a42889

Implement background music for ranked play (#37166 )

Using a bespoke music loop, fades in during the intro:


https://github.com/user-attachments/assets/8db7e30c-2a7c-4fd0-8385-fa77a6e0a506

Ducks when hovering cards for previews, with debouncing to smooth it:


https://github.com/user-attachments/assets/b3557c08-c675-4591-963c-95866d45107f

---
- [x] depends on ppy/osu-resources#414

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Jamie Taylor · 2026-04-02 17:12:25 +09:00

c4f2c66fd6

Update room state on card played in TestMultiplayerClient (#37172 )

Noticed this not updating when working on tests.

Krzysztof Gutkowski · 2026-04-02 16:19:02 +09:00

2d85966074

Fix multiplayer spectator player names being too small with new scrolling text (#37164 )

Addresses inline comment and
https://github.com/ppy/osu/discussions/37161.

Dean Herbert · 2026-04-02 08:06:42 +02:00

a8b66df9fb

Fix song select dropdowns not being navigable with keyboard (#37158 )

- Closes https://github.com/ppy/osu/issues/31684

Uses the global action variant of up and down for dropdown menus. It is
already used in multiplayer/playlist room/beatmap navigation and
gameplay menu overlays.

Comment wording is derived from:

https://github.com/ppy/osu/blob/fc817627e56b7db1c46e2f00fc9a3bd14af51211/osu.Game/Graphics/UserInterface/FocusedTextBox.cs#L73-L74

Joseph Madamba · 2026-04-01 20:59:47 +09:00

556be552c3

Fix replay/spectator scroll text not toggling with Ctrl+H (#37027 )

Intended to add toggle but forgot.

This also fixes https://github.com/ppy/osu/issues/37012 via a convoluted
refactor of a lot of stuff. The basic overview is:

- Moved all replay overlay concerns out of `HUDOverlay`. We can display
this above everything confidently (i think).
- Split out `ReplayOverlay` and `ReplaySettingsOverlay` so the base
class can handle the visibility, hotkeys and everything that should be
shared with *all* replay overlay components going forward. `Ctrl+H` is
supposed to hide any of these kinds of details, and I'm sure we'll add
more in the future.
- Reorganised some things in `Player` so the new structure would work.
Mainly the overlays which add a black layer during fade out.

Dean Herbert · 2026-04-01 11:29:25 +02:00

279effe23c

Ranked Play: Make cards draggable and reorderable (#37157 )

Adds the ability to drag and reorder cards. Card order is preserved
between rounds and is synchronized between both players (each player can
see the other player drag around and reorder their cards).

To make this possible I had to rewrite the card layout algorithm to be
stateless (e0a46fefaf), there wasn't
really a way around that since I needed a way to calculate a layout
position based on a card's index. Should hopefully be a lot easier to
read now though.

Some noteworthy stuff:
- I didn't really know what the best place to store the card order is,
so I put it on `RankedPlayCardWithPlaylistItem` since that one will stay
the same instance per round.
- To prevent the opponent's cards from dragged into the middle of the
playfield, only the x axis of the drag gets synchronized for the
`OpponentHandOfCards` with a fixed y value.
- I adjusted the replay recorder/player parameters a little. With the
drag events happening every frame the replay recorder record new frames
every 25ms and end up dropping half the replay frames per flush
interval, so I increased the sample interval to 50ms so the buffer size
matches the sample rate exactly (50ms -> 20 samples per flush every
1000ms).
I also increased the buffer size in the replay player a bit so slight
fluctuations in latency won't make it start to drop frames.


https://github.com/user-attachments/assets/b810cb85-db02-4edf-a63e-bfc96cf59665


https://github.com/user-attachments/assets/4d2f884d-fcce-4948-9659-fbb314634cb8

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

maarvin · 2026-04-01 18:21:44 +09:00

c4a49f647f

Fix editor not showing beatmap background in some cases

Dean Herbert · 2026-04-01 02:03:11 +09:00

2026.401.0-lazer 5c20254f76

Fix minor oversight in transform logic

Dean Herbert · 2026-04-01 01:14:02 +09:00

bb63a17c30

Fix regression in mania scroll speed calculation logic

Dean Herbert · 2026-04-01 01:03:00 +09:00

b3c15339d0

"Fix" performance drop caused by changing scroll speed (#37149 )

The fix is just disabling the animation. It works I guess.

---

- Closes https://github.com/ppy/osu/issues/37042

Currently in Mania, you can change the scroll speed for a brief period
during the beginning of a song. However this scroll speed change occurs
over a short period of time, which causes a bunch of extra hit object
updates, causing major fps and latency drops.

This fix simply replaces the dampening with an immediate scroll speed
update. Since the scroll speed can only be updated for a short time at
the beginning of the song, providing immediate visual feedback on the
scroll speed makes sense to me. However another potential solution would
be to filter the TimeRange Value updates to keep the gradual scroll
speed visual change, while greatly reducing the number of updates to the
hit objects currently on screen.

If there is any feedback I would greatly appreciate it as this is my
first issue here. I had ran both inspectCode.ps1 and the code formatter
before creating the merge request. Thank you.

Before fix:

https://github.com/user-attachments/assets/55e30894-7341-414a-af2e-2ec051c3a252

After fix:

https://github.com/user-attachments/assets/c085d33f-c0ae-45dd-8131-e79a5682b9ca

---------

Co-authored-by: Dean Herbert <pe@ppy.sh>

Austin Moore · 2026-03-31 23:33:53 +09:00

cbcbc788b8

Update resources

Dean Herbert · 2026-03-31 19:42:47 +09:00

3edc428c3a

76753 Commits