yserver 1.8.0

What's Changed

• refactor: move and regroup tests out of the largest source files (source layout cleanup, phases 0–1) by @joske in #227
• refactor: split process_request.rs into per-extension modules (source layout cleanup, phase 2) by @joske in #228
• refactor: split kms/render/backend.rs into portable and KMS modules (source layout cleanup, phase 2) by @joske in #229
• refactor: Backend trait impl forwards to per-subsystem helpers (source layout cleanup, phase 2.11b) by @joske in #230
• refactor: split kms/render/engine.rs by lifecycle and responsibility (source layout cleanup, phase 2) by @joske in #231
• refactor: split kms/render/scene.rs by compositor stage (source layout cleanup, phase 2) by @joske in #232
• refactor: split kms/render/platform.rs and kms/vk/scanout.rs into modules (source layout cleanup, phase 2) by @joske in #233
• fix(kms): copied PRIME transport — system memory and a sink-importable pitch by @ariel3259 in #235
• refactor: split XI and RANDR dispatch into per-request functions (phase 2c) by @joske in #236
• refactor: explicit imports instead of super-module globs by @joske in #237
• ci: run clippy and the tests on FreeBSD by @joske in #239
• refactor: move the Vulkan renderer into a yserver-render crate by @joske in #240
• refactor: move the scene compositor into yserver-render by @joske in #242
• ci: check the source layout rules by @joske in #243
• Hardware cursor in the wlroots model by @ariel3259 in #244
• fix(kms): atomic cursor follow-ups — failure attribution, plane assignment, hide after disable by @joske in #245
• fix(kms): give every CRTC a cursor plane when a full assignment exists by @joske in #246
• fix(input): key events carry window-relative coordinates and reach a selecting ancestor by @joske in #250
• feat(macos): run yserver in a window on macOS (Apple Silicon, KosmicKrisp) by @joske in #251
• fix(transport): turn Nagle off on TCP connections by @joske in #255
• Tearing on demand(C1): uncapped direct scanout for fullscreen Async clients by @ariel3259 in #249
• fix(xkb): apply modifier locks from XkbLatchLockState by @petrovo-as in #254
• feat(macos): Dock and app icon by @joske in #256
• fix: windows stay blank after a workspace switch under picom (#248) by @joske in #257
• fix: match Xorg's expose and damage events for ClearArea, compositing and MapSubwindows by @joske in #258
• fix(xkb): keep modifier locks across setxkbmap, send indicator and state notifies like Xorg (#253) by @joske in #259
• fix(composite): a redirected window's backing keeps its and its subwindows' borders by @SetZero in #260
• perf(core): keep a client's writer fd in the poller only while output waits by @SetZero in #262
• fix(composite): draw at the right offset in redirected windows, keep borders and children on resize by @joske in #265
• fix: window decorations smudged when resizing under a compositor (fvwm + xcompmgr) by @joske in #270
• fix(props): property data in server byte order, RotateProperties, and smaller property fixes by @joske in #271
• chore(deps): bump proptest from 1.11.0 to 1.12.0 by @dependabot[bot] in #269
• chore(deps): bump objc2 from 0.6.4 to 0.6.5 by @dependabot[bot] in #268
• Feat/drm commit identity by @ariel3259 in #272
• feat(macos): rootless mode — every X window its own Mac window by @joske in #274
• feat(kms): cursor sprite pool and atomic cursor state (2/5) by @ariel3259 in #276
• fix(kms): switching off or unplugging one monitor no longer blanks the others (#273) by @joske in #278
• fix(input): passive grabs report their activating event on the grab window (#277) by @joske in #283
• fix(kms): leave direct scanout on freshly composed frames (#281) by @joske in #284
• fix(kms): RandR SetCrtcConfig reconfigures only the named CRTC (#275) by @joske in #285
• ci: run the lavapipe suite on ubuntu-26.04 and fail when it lacks the scanout extensions by @joske in #286
• fix(kms): retirement cursor show uses the current pointer and newest image (3/5) by @ariel3259 in #287
• fix(kms): deferred cursor changes keep direct scanout, retried per output by @ariel3259 in #288
• fix(kms): direct scanout across monitors whose refresh rates differ slightly by @joske in #289
• feat(macos): a Settings window in yserver.app by @joske in #291
• fix(kms): the PRIME probe helper runs without DISPLAY by @ariel3259 in #292
• fix(kms): cursor errors only blame the cursor on EINVAL; remove the legacy cursor ioctls (5/5) by @ariel3259 in #290
• feat(kms): gamma through atomic GAMMA_LUT; remove the legacy SETGAMMA ioctl by @ariel3259 in #293
• feat(kms): VRR on the output a client window scans out directly (--vrr) by @ariel3259 in #294
• OpenBSD support by @joske in #296
• Fix a crash in reparent_subwindow after a window tree was half destroyed by @joske in #300
• fix(scanout): a shaped-away COW stage Present does not unflip direct scanout by @ariel3259 in #299
• Send PresentWindowDestroyed and the Ungrab focus/crossing events Xorg sends when windows and grabs go away by @joske in #301
• Selections are owned by the client that set them, as in Xorg by @joske in #302
• A second window manager is refused with BadAccess, as on Xorg by @joske in #303
• feat(kms): --vrr all for NVIDIA clients, honouring the NVIDIA and Mesa application lists by @ariel3259 in #298

New Contributors

• @petrovo-as made their first contribution in #254
• @SetZero made their first contribution in #260

Full Changelog: v1.7.1...v1.8.0