Skip to content

Latest commit

 

History

History
122 lines (88 loc) · 4.49 KB

File metadata and controls

122 lines (88 loc) · 4.49 KB

Stability And Support

This is a practical trust document for SeedBox as it exists in this repo. It explains what is well-proven, what is deterministic, what still needs bench validation, and what "supported" means here.

What is well-proven today

  • The shared core architecture is deliberate and repeatedly documented across firmware, native, and JUCE surfaces.
  • The native simulator/test lane is a strong proof surface for shared logic, scheduler behavior, UI snapshots, and deterministic render paths.
  • The golden pipeline is one of the repo’s clearest trust assets: fixtures, manifests, hashes, and control logs give reviewable receipts.
  • The build and hardware documentation is substantial enough to make SeedBox legible as a real builder ecosystem, not just an internal experiment.

What is deterministic and test-backed

These are the strongest "receipts" in the repository:

In practical terms, this means the repo has strong evidence for:

  • seed-driven shared logic
  • deterministic scheduler/pattern behavior
  • repeatable native renders and control ledgers
  • regression review based on artifacts, not memory

What still requires bench validation

Native proofs are not the same thing as full hardware proof. Bench validation is still required for:

  • exact hardware wiring outcomes
  • OLED, encoder, button, and MIDI physical behavior on a given unit
  • audio-shield noise/performance on a given bench supply and enclosure layout
  • timing and transport behavior when real external devices are involved
  • real-world parity checks between a physical build and the simulator/JUCE path

If you are building or demonstrating a physical unit, treat the following docs as mandatory companions, not optional reading:

What "supported" means here

In this repo, "supported" means some combination of the following is true:

  • there is an intentional code path in the repository
  • there is documentation for how to use or build it
  • there are tests, goldens, or explicit validation rituals around it
  • the maintainers are trying to keep the story coherent across surfaces

It does not mean:

  • guaranteed plug-and-play behavior on every machine or bench
  • complete polish across every UI and build path
  • hardware QA equivalent to a commercial manufacturing pipeline
  • perfect parity between all targets under all conditions

Simulator vs hardware parity

The right mental model is "shared musical logic, different physical realities."

What the simulator is good at proving:

  • the seed/app/scheduler logic
  • many UI state transitions
  • deterministic output and regression coverage
  • fast iteration without hardware friction

What hardware is still needed to prove:

  • physical IO correctness
  • analog and codec behavior
  • cable, power, and peripheral interactions
  • ergonomic truth at the panel

The simulator is not fake. It is one of the instrument’s main proof surfaces. But it should not be mistaken for a full substitute for bench validation.

Where experimental edges live

The main experimental edges currently appear around:

  • the broader JUCE desktop lane as a public-facing experience
  • hardware parity beyond documented build/bring-up rituals
  • future-facing roadmap areas that are intentionally visible in docs/roadmaps/
  • parts of the teaching surface that are strong but not yet consolidated into a single guided curriculum

What this repo promises

  • clear source access
  • documented build paths
  • visible caveats
  • deterministic proof surfaces where available
  • ethics and signing notes that make trust discussable

What this repo does not promise

  • fictional certainty
  • hidden maturity claims
  • consumer-product smoothness across every path
  • that an experimental lane should be read as a finished one

TODOs

  • TODO: add dated, unit-specific bench receipts once repeated calibration runs start landing in the docs tree.
  • TODO: add recurring dated JUCE smoke entries once actual host/runtime passes are being logged with the checklist.