From chatgp
Yes — there is one potentially important WebMIDI change around Chrome 152, even though neither Chrome 151 nor 152 lists a Web MIDI API change in the official release notes. �
Chrome for Developers +1
The significant one is macOS + Chrome 152. On July 22 Chromium substantially refactored its macOS MIDI backend to use the newer CoreMIDI MIDIEventList / Universal MIDI Packet (UMP) APIs, converting between UMP and the MIDI 1.0 messages exposed through Web MIDI. That commit is Chromium main position #1665988. �
Chromium
Chrome 152 branched later, at main position #1669021, so that new MIDI implementation is definitely part of Chrome 152. �
Chromium
And importantly: it introduced a real SysEx regression. Chromium subsequently found that SysEx7 conversion was wrong: the code incorrectly included F0/F7 inside UMP payloads and failed to reconstruct them correctly on incoming messages. Chromium explicitly describes this as:
“breaking Web MIDI SysEx on the default macOS UMP backend”
The fix landed August 3 at main position #1672730. �
Chromium
That position is after the M152 branch point, and I can’t find evidence that the fix was merged/cherry-picked into the 152 branch. So if your problem is Chrome 152 on macOS and involves SysEx, I would consider Chrome 152 itself a very strong suspect.
For Chrome 151, there wasn’t a comparable public WebMIDI API change. There was some internal Windows work before the M151 branch, including a June 4 fix to the experimental WinRT MIDI backend concerning stale callbacks/session rebinding. � The WinRT backend is still disabled by default in Chromium, though. �
Chromium
Chromium
So the short version is:
Chrome 151: no obvious intentional WebMIDI behavior change.
Chrome 152 / macOS: yes, major backend change, plus a known SysEx regression.
Chrome 152 / Windows: I haven’t found a comparable change that should alter ordinary WebMIDI behavior.
The proposed change making MIDIAccess transferable/available in workers is only an Intent to Prototype from August 25, not something that shipped in 151/152. �
Mail Archive
If you’re seeing something break specifically between 151 → 152, tell me OS + what exactly fails (device enumeration, MIDI input, output, SysEx, reconnect, etc.) and I can trace the exact Chromium changes involved.
I can also keep an eye on whether the SysEx fix gets backported in to Chrome 152 stable.