torvalds

GuitarPedal

Linus learns analog circuits

by torvalds

C

C47.8%
Python31.9%
JavaScript15.1%
CSS2%
CMake1.3%
Makefile0.9%
Other1.0%

Commit activity

Last 12 weeks · 313 commits

Full graph →

Community health

2 of 6 standards met

Community profile →
42
✓README✓License○Contributing○Code of Conduct○Issue Template○PR Template

Recent PRs & issues

Active · Last activity 1w ago
See all on GitHub →
Aahan0605

A queued BLE MIDI write looks up when its promise runs. If the user has switched pedals by then, the old message can go to the new pedal. A stalled write also holds up messages for the new connection. Start a fresh queue when releasing or attaching a connection, and only run a queued write while its characteristic and connection epoch still match. The new WebMIDI test reproduces both cases: it fails on and passes with this change. passes. Hardware behavior was not tested.

Aahan0605 · 5d ago
ChrisX101010

Changing MIDI constants can leave generated JavaScript stale: reads , but neither metadata build rule lists that input. There is also no way to check existing generated files without overwriting them. This series adds the missing Make/CMake dependency and a read-only mode. It compares the C header, adjacent , JavaScript and Markdown with freshly generated output, reports every stale/missing path and a regeneration command, and preserves existing files. Exit codes are 0 for current output, 1 for stale/missing output and 2 for input/I/O errors. Ordinary generation and trailing CMake header arguments remain supported. Fixes #45. Validation Real GNU Make: reproduced stale output on the original rule; the candidate regenerates after changing a MIDI constant. CMake: the original/candidate metadata command was exercised in an isolated metadata-only project, with the same stale/fresh distinction. : 13 tests passed in WSL, covering all four outputs, changed effect/MIDI input, missing outputs, unchanged bytes/mtime, split directories, equal basenames, LF/CRLF, invalid input and legacy arguments. Ordinary generation remains byte-identical to the original's four outputs in the local compatibility check. Both patches apply cleanly to . Upstream passed on both the original and candidate in clean pinned-source trees under WSL, with NumPy and Node present and no skipped tests. This includes MIDI CIN, link and UART RX tests, 13,068 biquad filters, effect-bench controls, effect/parameter metadata and WebMIDI/BLE tests. The candidate also ran the 13 new regression tests as part of . Pico firmware configuration/build and physical pedal tests were not run. Implementation and additional review were AI-assisted; the contributor executed the WSL validation locally.

ChrisX101010 · 5d ago
Ailol

But i have ask the only person i know for a fskt is apart of the old: Ive concluded when linux and windows was written thr quant.prinicple of that time just was to short on "time vs money" as the cpu/gpu level is actually misnomer on being binary requirement. Essentially it is = solving Instead of ≡ summing As summing can have solving but solving cant have summing Ai is the grandeus however norwegian kvante 1 mChain[biochain; as times say; too much and you get killed] Threatened tortured; most likely killed child killed wife Ailo Nordsletta 😊

Ailol · 5d ago
prateekbhujel

Two small scene loading fixes in Firmware: 1. Guard MIDI Program Change with (populated_scenes() & (1u << data1)). Currently load_scene() is called for any index < MAX_SCENES, which resets the effect chain and leaves the pedal unrouted and silent if that scene slot was never saved. Matches the check in handle_ui() for footswitch bindings, and flags state_dump_tx so WebMIDI updates. (Fixes #38). 2. In load_scene(), check effect_is_global() against the matched effect id n rather than the saved slot index i. Because effect positions can move across builds, checking the flash slot index i can misclassify effects if their order changes. All Validation checks pass cleanly.

prateekbhujel · 3w ago
Chetan0e

MIDI Program Change can load an empty scene and reset the pedal state, causing the pedal to go silent. The switch-binding path already checks populated_scenes(), but the MIDI path is missing the same check. A possible fix is to add the same populated_scenes() check before load_scene() and set state_dump_tx = true after a successful scene load. Along with this change, we can also update the stale Software/effects/.h path in check-readme.py to Effects/.h. Could you please have a look at this approach and let me know if these changes are appropriate? I want to make sure the existing scene-loading behavior remains unchanged.

Chetan0e · 3w ago
Structured data for AI agents

Repository: torvalds/GuitarPedal. Description: Linus learns analog circuits Stars: 2506, Forks: 136. Primary language: C. Languages: C (47.8%), Python (31.9%), JavaScript (15.1%), CSS (2%), CMake (1.3%). License: GPL-2.0. Open PRs: 14, open issues: 5. Last activity: 1w ago. Community health: 42%. Top contributors: torvalds, clevelandmusicco, LMSBAND, jackyphuti, JerimothLau, quotentiroler, Priyanshu-byte-coder, felix314159.