Repository navigation
memory cues: create from the web UI, survive the wire - #48
Merged
Merged
Conversation
Issue #46: memory cues imported from rekordbox served and displayed fine, but the web UI could only create hot cues - + CUE picks the lowest free pad in A-H and there was no creation path for the 9+ range at all. The server API accepts any number, so this is UI-only. A + MEM button next to + CUE adds a memory cue at the cursor, numbered from 9 up (matching how imported rekordbox memory cues are stored, so serving, exports and the cue list treat both kinds identically). Default colour, like the deck's own memory cues. Verified headless: + CUE then + MEM twice yields cues 1, 9, 10 via the API.
Field report: setting a second memory cue from the CDJ overwrote the first. Decks mark memory cues with 0 in the cue blob's number field - that's the wire convention separating them from hot pads 1-8 - and the store keys cues by number, so every deck-saved memory cue landed on key 0. Verified against a capture of real rekordbox taking two memory-cue saves from a CDJ-2000NXS2: both 0x2705 blobs carry number 0, and rekordbox's 0x4e02 responses serve them back with number 0. Inbound: a 0x2705 cue with number 0 is remapped to the next free internal memory number (9+, the convention imported rekordbox memory cues already use); the raw blob is stored as-is since it already says 0, which is what the deck must be served back. Outbound: MarshalCueBlob now writes 0 on the wire for internal numbers 9+, so synthesized memory cues (web-created or imported without raw blobs) stop going out as nonexistent hot pads. Also drops MarshalCueBlob's stale doc block (described a 76-byte layout; the function has emitted 124-byte NXS2 blobs for ages). TestDeckMemoryCueSavesDontOverwrite and TestMarshalCueBlobMemoryWireNumber pin both directions.
vynulldev
changed the title
vynulldev
marked this pull request as ready for review
October 3, 2026 20:27
5 tasks done
vynulldev
added a commit
that referenced
this pull request
Oct 3, 2026
Behavior-neutral v0.5.0 audit. Test hygiene: the real-audio beat and phrase harnesses move from link/prolink (where the part-3c encoder relocation stranded them) to analysis/, switching to the format-aware encoder delay on the way (they called the lossy default, shifting lossless files' printed grids by 25ms); encoder dumps stay in prolink as encoders_realaudio_test.go; bench_test.go becomes pipeline_timing_test.go; proto/dbmessage_test.go drops two tests skipped since the initial commit (a marshal->parse combination that doesn't exist on the wire) and notes that real coverage needs golden fixtures captured from a deck session. go mod tidy was a no-op. README: serve-a-rekordbox-USB (the release headline) gets a feature bullet and a rewritten section (default-mode example without sudo, what decks/web get, read-only rule, OneLibrary dual-format note); the colour-waveform bullet no longer claims FFT analysis (it's been time-domain Butterworth banks for ages); the BPM/beat-grid bullets now describe the shipped tempogram engine; the cue bullet covers hot and memory cues (post-#48).
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
1 participant
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.
What & why
Shoutout to @kayrozen (#46) for the report!
Two halves, one subject: memory cues were second-class citizens, creatable nowhere and mishandled on the wire in both directions.
Web UI (1c37a29), fixes #46: memory cues imported from a rekordbox library displayed fine, but the UI could only create hot cues.
+ CUEpicks the lowest free pad in A-H, and the 9+ memory range had no creation path. A+ MEMbutton next to+ CUEadds a memory cue at the zoom-waveform cursor, numbered from 9 upward (the convention imported rekordbox memory cues already use, so the cue list, exports and serving treat created and imported cues identically). Verified headless with Playwright:+ CUEthen + MEM twice yields cues 1, 9, 10 through the API.dbserver (50183da): setting a second memory cue FROM THE DECK overwrote the first. Decks mark memory cues with 0 in the cue blob's number field: that is the wire convention separating them from hot pads 1-8, and the store keys cues by number, so every deck-saved memory cue landed on key 0. Confirmed against a packet capture of real rekordbox taking two memory-cue saves from a CDJ-2000NXS2: number 0 inbound on both saves, and rekordbox serves them back with number 0 too.
The fix translates the convention at the wire boundary, both directions. Inbound, the 0x2705 handler remaps number 0 to the next free internal memory number (the raw blob is stored as-is, it already says 0, which is exactly what the deck must be served back). Outbound, MarshalCueBlob writes 0 on the wire for internal numbers 9+, which also fixes a latent bug: memory cues imported from rekordbox libraries had been served to decks as nonexistent hot pads 9+ since #42.
Bonus RE from the capture, recorded for later: rekordbox's 0x4e02 responses use compact 56-byte cue records; our 76/124-byte records are NXS2-proven but worth normalizing if other models misbehave. The deck-side memory-cue DELETE message was not in this capture and remains unsupported until one is.
TestDeckMemoryCueSavesDontOverwrite and TestMarshalCueBlobMemoryWireNumber pin both directions.
Hardware testing
This changes what decks receive for memory cues (the wire number mapping) and how deck saves are stored. The v0.5.0 checklist covers both directions on the rig: two deck-saved memory cues persist and come back on re-load, and a web-created memory cue renders as a MEMORY cue (not a hot pad) on the CDJ.
Checklist
go build ./...,go vet ./..., andgo test ./...passgofmt -l .is cleanGPL-3.0-or-later) (n/a, no new files)