Repository navigation
Release Gates
Since 2026-10-26 the rules for the next releases are machine-readable (docs/status/release.yaml in OpenIMP next) and the generated feature matrix shows their state. The user decision behind them: a release is not a date, it is a state that is measured.
Where to look:
-
Generated matrix with the gate section: https://lu-fi.github.io/openimp/feature-matrix.html (hover or click a cell for state, evidence and issue) and
docs/FEATURE_MATRIX.mdin the repository. -
Project board: https://github.com/users/Lu-Fi/projects/1 (gaps, bugs and progress are public GitHub issues in Lu-Fi/openimp; milestones RC3 and Release 1; an open issue labelled
type:wipis a construction sign in the matrix). Tracking issue: #107. -
Rule files:
docs/status/release.yaml(gates),docs/status/features.yaml(45 features x SoCs),docs/status/issues.json,docs/status/soak.json;make status-checkvalidates them.
T10, T20, T21, T23, T31, T40, T41. T30 (no camera), T32 and T33 are not part of RC3 or Release 1.
Every vendor function that a streamer (timps, prudynt, raptor) calls works on every gate SoC. "Works" means the API matrix cell is dev, dev+, run, rt or a vendor no-op that OpenIMP matches (see Testing for the evidence levels). The streamer call sets are the "Used by" data of docs/api_evidence.json; fleet runs are in docs/api_run.json.
State of next on 2026-10-27 (generated): 1787 of 1787 streamer-function cells work, 0 missing: the RC3 gate is GREEN.
| SoC | works | missing |
|---|---|---|
| T10 | 228/228 | 0 |
| T20 | 228/228 | 0 |
| T21 | 252/252 | 0 |
| T23 | 290/290 | 0 |
| T31 | 319/319 | 0 |
| T40 | 217/217 | 0 |
| T41 | 253/253 | 0 |
RC3 is not tagged yet. Next steps: build an RC3 candidate image, run it for one night of soak, then tag RC3. Some cells count as working at the evidence level "API round trip" (for example DMIC_* on T31, T40 and T41) or as a matched vendor no-op; the matrix shows which. The ISP bypass is one of the matched cases: on T10, T20, T21, T23 and T31 SetISPBypass(ENABLE) (full ISP, the streamer's start-up call) is a no-op returning 0 and DISABLE (the real bypass) is refused with -1 like the vendor T21 library (issues 119 and 126); on T41 Set 0 works and Set 1 is refused with -1 (issue 125). A working real bypass is issue 129.
RC3 is green (reached) and published (tag), and the image built from that tag then runs for at least 14 days without a reset on the fleet:
- at least 1 camera per gate SoC reaches the 14 stable days (a SoC is as good as its best camera),
- 0 unplanned reboots and 0 kernel oopses in the counted window,
- a reset is an unplanned reboot, an oops or a new image, and restarts the counter,
- planned flashes and tests are excluded only if the camera's status log marks the line as "planned".
Release 1 = RC3 tag + 14 days on the fleet. The 14-day counter starts only when RC3 is published. State on 2026-10-27: not started (0 of 7 SoCs). The soak data is refreshed daily (make status-soak).
Run the device test app on a camera of your SoC (Testing) and attach apitest.json to an issue: a PASS on a camera turns a cell from "implemented, not device-tested" into evidence.
Open Ingenic camera stack
- Home
- Supported SoCs and Sensors
- Per-SoC Status
- Feature Matrix
- Architecture
- Building
- Installing and Switching
- Module Parameters
- Environment Variables
- Beyond the Vendor
- API reference
- Memory
- Troubleshooting
- Testing
- Test Pattern
- Performance
- Changelog
- Release Scheme
- Release Gates
- Live feature matrix
- Project board
Current: RC2 (v2026.10.25-rc2); RC3 in progress