Bug 498027

Summary:	kwin ignores applications' initial geometry requests for native Wayland windows
Product:	[Plasma] kwin	Reporter:	Oswald Buddenhagen <ossi>
Component:	wayland-generic	Assignee:	KWin default assignee <kwin-bugs-null>
Status:	CLOSED UPSTREAM
Severity:	normal	CC:	kde, kdedev, nate
Priority:	NOR
Version:	6.2.4
Target Milestone:	---
Platform:	Debian unstable
OS:	Linux
Latest Commit:		Version Fixed In:
Sentry Crash Report:

Description Oswald Buddenhagen 2024-12-29 16:59:33 UTC

kwin seems to have become extremely opinionated about window placement in plasma 6.
specifically, it seems to ignore the position specified on the command line. given this command:

  qdbusviewer6 --qwindowgeometry 400x800+900+100

the window size is heeded, but the position is plainly ignored.
interestingly enough, the following command does what it should:

  xmessage -geometry 120x80+900+900 foo bar

so as a first guess i'd say it's only ignoring the netwm hints, but heeds the old-style x11 geometry.

KDE Frameworks Version: 6.8.0
Qt Version: 6.7.2

Comment 1 David Edmundson 2025-01-02 15:20:38 UTC

`  qdbusviewer6 --qwindowgeometry 400x800+900+100 --platform xcb`

 works as expected

With Wayland it is expected that clients cannot control placement at a protocol level.

Comment 2 Oswald Buddenhagen 2025-01-03 00:58:55 UTC

so yet another intentional regression. truly the way of the future.

but i'm stunned that i actually _am_ on wayland. something went wrong with session selection in lightdm, and i didn't even notice, except for a few regressions.

Comment 3 Oswald Buddenhagen 2025-02-01 16:05:17 UTC

*** Bug 498030 has been marked as a duplicate of this bug. ***

Comment 4 Oswald Buddenhagen 2025-02-01 16:13:32 UTC

let me elaborate:
- this breaks scripted interactions like via kdialog (which was my original reason to make this report)
- it breaks window placement upon session restore (whether by the session manager or the apps themselves)
- it produces weird effects like bug #498027

so to put it into absolutely unambiguous terms: this IS a bug, and anyone who claims otherwise needs to recheck their priorities.

i don't care if the protocol designers had /reasons/ to make it that way, for example because of security. design something that doesn't suck for users.

Comment 5 Nate Graham 2025-02-03 22:33:13 UTC

FWIW I share your frustration at the situation, but at least there's a solution for #2 on the horizon in the form of a new Wayland session restore protocol. Once it's finalized, we'll support it.

Regardless, bugs are where something doesn't work as designed; unfortunately this does work as designed, as it's a core part of Wayland that the window manager alone is in charge of positioning for native Wayland windows, with limited exceptions. This bug report isn't the place to argue against that; it is what it is.

Comment 6 Oswald Buddenhagen 2025-02-04 08:44:42 UTC

> bugs are where something doesn't work as designed

bad designs are also bugs, and it's nonsensical to argue against that.

> as it's a core part of Wayland that the window manager alone is in charge of positioning for native Wayland windows

yeah, because giving control to primitive machines has always been such a great idea. totally.

please reopen the bug, and link the upstream dependency appropriately.