Repository navigation
Push opens on the RC-500 into an empty memory, and nowhere else - #174
Open
AliceLafox wants to merge 2 commits into
Open
AliceLafox wants to merge 2 commits into
AliceLafox wants to merge 2 commits into
Conversation
Decided 2026-10-29: upload on the two-track model for the one case with no unknowns - a memory with no take on either track - and a refusal in a sentence for the rest, until the pedal has been asked what it makes of a second take of another length beside a pushed one (hardware reading 2026-09-23: a later track 2 came out exactly half of track 1; a non-multiple length is unknown). kRc500 opens push. In commands::push the memory is read through the gate on every path, the drop-only one included, before a byte moves; on a model with more tracks than one, a take on any track - configured or merely in its folder - refuses the push naming that track, force or not. The take lands on track 1 with the body the one-track model gets: TRACK1 WavStat, WavLen, MeasLen, Measure, RecTmp; MASTER Tempo and LpLen, which the pedal keeps as the length of the first recording (2026-09-23), and the pushed take is that recording. CurrentTrack and LpMod stay as the memory has them: single-track recordings off the pedal carry CurrentTrack 0, the factory value, and LpMod 0 on one such memory and 1 on another (2026-09-22), so neither is the recording's doing. The app follows the rule through catalog::uploadOffered: the slot menu's push item, the drag's highlight and the drop land only where an upload is offered; the RC-500's "closed" sentence lists push among the open operations. A push into an empty two-track memory is recorded on track 1 as an RC-5 push is, and a restore of the state before it leaves both tracks empty. rc500_push_tests, from the spec: the empty memory takes the upload with the fields above and track 2's folder untouched; track 1 occupied, track 2 alone, a stray file in a folder - each refused naming the track, the card byte for byte as it was; the drop-only path refused by the same rule and, on a card the gate does not read, before the file lands; the restore. slot_table_harness: a drop on the track-2-only memory calls nobody, on the empty one lands, on an occupied RC-5 slot is a replace. Two mutants against the suite, each caught: the occupancy rule dropped, and the drop-only path reading nothing.
…t asks every track The review found the app offering a push the core refuses: the rule in the app read the config alone (catalog::uploadOffered), while push also looks into the track folders, so a two-track memory with a stray file and an empty config was highlighted, dropped on and listed as a target, and refused only in the toast. offersUpload (PedalWorker.h) now joins the config's say with the folders as the scan listed them, and the menu, the drag's highlight, the drop and the hint all ask it. The empty-slot hint read the flat TRACK1 field: a memory recorded on track 2 alone invited a drop it would not take and a click the core refused; it asks every track now. The greyed Replace item on a used two-track memory says why in its own label, since push is open on the model and the "closed" sentence is for closed operations. The harness presses the hint on the track-2-only and the empty memory, drops on a folder-occupied one, and checks that each drop point maps to its row. Not changed, by the review's reading: the RC-5 path is what it was, byte for byte; the two-track body writes land in TRACK1 and MASTER only; a push into an empty two-track memory records and undoes both tracks. Left open: no model in the table closes push any more, so the drop-only path's profile gate is proven with a card the gate refuses to read, not with a model that closes push.
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.
Decided
2026-10-29: upload on the two-track model for the one case with no unknowns, a memory with no take on either track, and a refusal in a sentence for the rest, until the pedal has been asked what it makes of a second take of another length beside a pushed one (hardware reading 2026-09-23: a track 2 recorded after track 1 came out exactly half its length; a non-multiple length is unknown).
The change
kRc500openspush. The RC-500 sentence becomesOn the RC-500, LooperCat does rename, set one-shot, push, pull, clear and restore only.commands::pushreads the memory through the profile gate on every path, the drop-only one included, before a byte moves. On a model with more tracks than one, a take on any track, configured or merely in its folder, refuses the push:push refused: memory 11 holds a take on track 2 - upload into a used "RC-500" memory is not offered yet. Force changes nothing there. The take lands on track 1 with the body the one-track model gets: TRACK1WavStat,WavLen,MeasLen,Measure,RecTmp; MASTERTempoandLpLen. Track 2 is untouched.LpLen,CurrentTrack,LpMod, from the pedal's own files (2026-09-22/23 dumps):LpLenis the length of the memory's first recording (a memory recorded track 1 first at 8 bars then track 2 at 4 carries 8; one recorded on track 2 alone carries that track's length), and the pushed take is that recording, soLpLen= its measures.CurrentTrackis 0 on every single-track recording and on an untouched memory, zero-based (1 after a recording on track 2), so an empty memory already carries the value a push on track 1 would write: left as it is.LpModis 0 on one single-track recording and 1 on another, so it is not the recording's doing: left as it is, and what it means is a checkpoint question.offersUpload(PedalWorker.h): the slot menu's item, the drag's highlight, the drop and the empty-slot hint land only where the core takes an upload, the config's say and the folders as the scan listed them. A greyed Replace item on a used two-track memory says why in its label.Tests
tests/rc500_push_tests.cpp, from the spec: the catalog's offer per memory; the empty memory takes the upload with the fields above, track 2's folder and section and the MASTER fields the pedal keeps for itself untouched, both banks alike; track 1 occupied, track 2 alone, a stray file in a folder, the memory just pushed, each refused naming the track, force or not, the card byte for byte as it was and no archive opened; the drop-only path refused by the same rule, and on a card the gate does not read, before the file lands; the restore.slot_table_harness: the permission table and sentence; the drag wanted on the RC-500; a drop on the track-2-only memory calls nobody, on the empty one lands, on a folder-occupied one calls nobody, on an occupied RC-5 slot is a replace; the hint's click on the track-2-only memory opens nothing, on the empty one opens the chooser. Four other suites moved with the sentence.Mutants against
rc500_push_tests, each caught: the occupancy rule dropped (10 failures); the drop-only path reading nothing (5 failures).Local: 85 of 85 suites and 8 of 8 harnesses pass on macOS, UTF-8 lint included.
Known
No model in the table closes push any more, so the drop-only path's profile gate is proven with a card the gate refuses to read (an unknown root name), not with a model that closes push.
Hardware check to close
At the RC-500: push one file into an empty memory; the pedal plays it and shows the same bars and tempo as the row. Then record track 2 on the pedal over it, at a length that is not a multiple of the first, and read the memory's body back from the card:
LpLen(still the pushed take's measures?),CurrentTrack(1 after the recording on track 2?),LpMod, andTRACK2/WavLenagainstTRACK1/WavLen. That reading, with a checksum of the pushed file before and after the pedal's boot, is the fact for push into a used memory, which stays closed until it is in.