Repository navigation
Highlights
- Display section is now live and accurate on Apple Silicon. Reference Mode (the user's chosen preset, e.g. "Apple XDR Display (P3-1600 nits)" or "HDR Video"), refresh rate / ProMotion, current/peak nits, and an HDR-active indicator are all sourced from in-process IOReport / CoreGraphics / CoreDisplay APIs. They update in real time as you move the brightness slider or change presets in System Settings β no subprocess, no Python.
- Mac Neo (M5 Pro/Max generation) display power is now read correctly. macOS introduced a new SMC key
PBwofor backlight power on these machines while leaving the legacyPDBReither dead or stuck around 0.3 W. macpow now prefersPBwoand falls back toPDBRon older XDR Macs (M1βM4). - New
--dump-smcdiagnostic flag β dumps every SMC key, type, and decoded value in the same format asdkorunic/iSMC raw, so users on unfamiliar hardware can attach a triage dump to issues.
Fixes
#9 β Python app flashing in Dock every few seconds
The periodic Python+ctypes subprocess that read NSScreen's EDR headroom registered as a GUI app and made a "Python" icon flash in the Dock every ~5 s. Even after avoiding NSApplication.sharedApplication and forcing setActivationPolicy:Prohibited, simply loading AppKit was enough to trip the Dock on some macOS versions. The integral fix: drop the AppKit subprocess entirely and source the same data from APIs that already publish it live in-process:
- Reference Mode β
CoreDisplay_Display_GetActivePresetIndex+_CopyPreset - HDR-active β IOReport
backlight report / DPB factorchannel - Refresh rate / ProMotion β
CGDisplayCopyDisplayMode/CGDisplayCopyAllDisplayModes
Also drops the runtime python3 dependency that came with it. Thanks @ylorn for the precise bug report and the trace.
#12 β Display power shows incorrect on Mac Neo
The fix matches the issue author's analysis: try the new PBwo key first, fall back to PDBR. Verified on M5 Max where PBwo is absent (so PDBR continues to be used) and on M5 Pro/Max/Neo where PBwo is the rail that actually tracks brightness. Thanks @jacobsraspberry.
While poking at SMC for this issue we also found and fixed a latent bug in SmcConnection::read_key_raw: macOS returns success with data_size=0 for keys that don't exist on the current SoC, which read_f32 silently surfaced as Ok(0.0) and broke any or_else fallback. Now those reads cleanly error out so callers can fall back deterministically.
New features
macpow --dump-smcβ full SMC key dump (name, type, decoded value, raw bytes), output format matchesiSMC rawso existing dumps indkorunic/iSMC/reports/are directly comparable.- Display label now reads e.g.
14" Apple XDR Display 1600/1600 nits 120Hz(withΒ· HDR x10appended when HDR content is on screen). The "max" reflects the peak the active Reference Mode can reach (e.g. 1600 in P3-1600, 600 in P3-600).
Internal
- Pinned toolchain via
rust-toolchain.tomlso localcargo clippy/cargo fmtuse exactly the same Rust version as CI'sdtolnay/rust-toolchain@stable. This prevents a class of release failures where a clippy lint promoted in a newer stable trips CI aftercargo publishhas already locked the version. - Tightened the Release Checklist in README.md accordingly:
rustup update stablefirst, copy the clippy command verbatim from CI, dry-run the publish before tagging, and don't force-push tags after acargo publish.
Compatibility
No breaking changes for end users. Drops the runtime dependency on python3 (no longer required at any point).
Full Changelog: v0.1.17...v0.1.19