Repository navigation
midi: a MIDI input chooses how its controllers are translated, and a pending RPN no longer silences it - #990
Open
jcelerier wants to merge 1 commit into
Conversation
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
had a problem deploying
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Failure
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 01:59 — with
GitHub Actions
Inactive
…pending RPN no longer silences it libremidi's stream translator replaces the whole-buffer one midi_protocol used per message: a keyboard sending its pitch bend range as CC 101, 100 and 6 without CC 38 left an RPN pending, after which every message of the device was dropped before reaching the graph, while the device tree still showed them. midi_protocol_configuration::controllers says whether controllers reach the graph as they were sent (raw, the default) or with RPN, NRPN and bank select combined into MIDI 2 messages (semantic). A message the translator cannot handle costs that message only, and is logged once per kind of error for the device. libremidi is moved to its stream translator, on top of master. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KdPgugz48pdcaqHJAg2vid
jcelerier
force-pushed
the
feature/midi-controller-translation
branch
from
November 20, 2026 15:45
5fd2d39 to
7f0079d
Compare
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Inactive
jcelerier
had a problem deploying
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Failure
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Inactive
jcelerier
temporarily deployed
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Inactive
jcelerier
requested a deployment
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Queued
jcelerier
requested a deployment
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
In progress
jcelerier
requested a deployment
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Queued
jcelerier
requested a deployment
to
Apple Certificate
November 20, 2026 15:45 — with
GitHub Actions
Queued
This branch is being deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #981 (base
feature/midi-note-off-cabls), and needs celtera/libremidi#295, whose branch the libremidi submodule points to.Problem
midi_protocol::midi_callbackconverted each incoming MIDI 1 message with libremidi'smidi1_to_midi2, which kept a global RPN / data entry state and failed every call while an RPN was pending. The callback ignored that result. A keyboard sending its pitch bend range asCC 101, CC 100, CC 6(no CC 38) therefore made every later message of the device vanish before it reached the graph, while the device explorer, fed further down the same callback, still showed the input. Every inlet reading the device went quiet. This was reproduced on the real ALSA sequencer path with the keyboard's exact sequence.Change
midi_protocol_configuration::controllerschooses between:raw(default): controllers reach the graph as they were sent;semantic: RPN, NRPN and bank select are combined into MIDI 2 messages.Verification
The probe used the real
midi_protocol, two devices on two virtual ALSA sequencer ports, and the keyboard's sequence repeated, then notes. In both modes, 300 of 300 later note-ons were queued for the graph. Raw mode delivers the three CCs as three CCs; semantic mode delivers one RPN on CC 6 and another on a later CC 38. score'stest_unit_midi_controller_translationcovers the same path throughmidi_protocol.🤖 Generated with Claude Code
https://claude.ai/code/session_01KdPgugz48pdcaqHJAg2vid