Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
27 commits
Select commit Hold shift + click to select a range
94a472c
feat: refresh CorOS 4.1 catalog and observations
tony-xmelon Sep 3, 2026
9f4910f
Address review feedback on firmware catalog
tony-xmelon Sep 3, 2026
a0c12ba
Merge upstream main into CorOS 4.1 catalog draft
tony-xmelon Sep 3, 2026
737c95e
Record CorOS 4.1 handshake behavior
Sep 4, 2026
0c9453f
Merge remote-tracking branch 'upstream/main' into codex/coros-4-1-cat…
tony-xmelon Sep 5, 2026
5ddd1fb
Merge CorOS 4.1 catalog into firmware profiles
tony-xmelon Sep 7, 2026
6885db0
Merge post-profile hardware records into CorOS 4.1
tony-xmelon Sep 8, 2026
fc5a24b
Refresh CorOS 4.1 parameter snapshot metadata
tony-xmelon Sep 8, 2026
f6b4b9b
Record verified CorOS 4.1 operations
tony-xmelon Sep 8, 2026
f25caf1
Correct CorOS 4.1 hardware observations
tony-xmelon Sep 11, 2026
cdccff7
Classify delayed Version announce replies
tony-xmelon Sep 11, 2026
5bfec83
Update profile connection expectation
tony-xmelon Sep 11, 2026
9b77102
Address CorOS 4.1 catalog review
tony-xmelon Sep 11, 2026
0b2e1a8
Merge remote-tracking branch 'upstream/main' into codex/coros-4-1-cat…
tony-xmelon Sep 12, 2026
b1baa5f
Merge remote-tracking branch 'upstream/main' into codex/coros-4-1-cat…
tony-xmelon Sep 19, 2026
76110fd
Follow documentation typography rules
tony-xmelon Sep 19, 2026
1a1da54
Merge remote-tracking branch 'upstream/main' into codex/coros-4-1-cat…
tony-xmelon Sep 22, 2026
9bf85a3
Record complete CorOS 4.1 connect observation
tony-xmelon Sep 22, 2026
6e98555
Address CorOS 4.1 review details
tony-xmelon Sep 22, 2026
db257b5
Merge remote-tracking branch 'upstream/main' into codex/coros-4-1-cat…
tony-xmelon Sep 22, 2026
283f00c
Refresh CorOS 4.1 catalog from current device
tony-xmelon Sep 22, 2026
df15e9b
Add CorOS 4.1 option evidence
tony-xmelon Sep 22, 2026
73350ff
Correct CorOS 4.1 evidence record
tony-xmelon Sep 26, 2026
9b3a96b
Address latest CorOS 4.1 review findings
tony-xmelon Sep 26, 2026
9f33e90
Clarify the 4.1 verified operation count
tony-xmelon Sep 27, 2026
8f09d5a
Pin CorOS 4.1 catalog counts and handshake evidence
tony-xmelon Sep 27, 2026
546b444
Merge current main into CorOS 4.1 catalog branch
tony-xmelon Sep 27, 2026
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 40 additions & 3 deletions changelog.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,39 @@ sustained stretch without a correction.

## Unreleased

### CorOS 4.1 catalog and measured profile

`QuadCortex41` now binds a generated CorOS 4.1.0 snapshot containing 422 factory
models and its firmware-specific parameter and option constants. Contributor
hardware runs on 2026-09-11 verified 16 inherited operations. The later #62
profile change removed `read_current_preset` from this profile's `VERIFIED`
set until that dependency lands, leaving 15 inherited operations currently
verified. The independently captured local-backup operation and contributed
device-name round trip are the sixteenth and seventeenth verified methods.
Unversioned `protocol.models`, `params`, and `options` remain the CorOS 4.0.1
compatibility snapshot.

The live CorOS 4.1.0 ModelRepo read on 2026-09-27 contains 633 models, 600
placeable models, and 398 models with resource metadata. Its `displayPos`
population is also different from 4.0.1; these are catalog counts, not new
screen readings. See `docs/domain-model.md` for the side-by-side counts.

The snapshots are intentionally distinct. Notably, Minivoicer mode value 2 is
`CHROM` on 4.0.1 but `NATURAL_MINOR` on 4.1.0, and the 4.1 Overlord Synth scale
is no longer a boolean-shaped control. Callers should use `qc.models`,
`qc.params`, and `qc.options` from the connected profile.

Six factory names also changed at stable wire IDs, so code using the old names
must select names from the connected profile:

| ID | CorOS 4.0.1 | CorOS 4.1.0 |
|---:|---|---|
| 3000 | `MICROTUBES_B3K` | `DOUGLAS_MT_3K` |
| 3006 | `MICROTUBES_VMT` | `DOUGLAS_VINTAGE_MT` |
| 21001 | `N210C_DARKGLASS_M` | `N210_DOUGLAS_CERAMIC_M` |
| 21005 | `N212_DARKGLASS_NEO_M` | `N212_DOUGLAS_NEODYMIUM_M` |
| 33001 | `N210C_DARKGLASS_ST` | `N210_DOUGLAS_CERAMIC_ST` |
| 33005 | `N212_DARKGLASS_NEO_ST` | `N212_DOUGLAS_NEODYMIUM_ST` |
### Rename the unit, drive undo/redo, and read inhibited modules

`set_device_name()` sends a sparse Version update. `undo()` and `redo()` drive
Expand Down Expand Up @@ -44,9 +77,13 @@ all seven out, read on the unit 2026-09-16 with positions 0 and 3 driven from a
host. The same word draws two ways on two controls, so nothing in the catalog's
text predicts which.

Nineteen controls have now been read. Two shorten a word, four draw circles
Twenty controls have now been read. Three shorten a word, four draw circles
instead of words, and the other thirteen match the catalog. `Osc1Wave` is still
the only enum whose names are known to differ from what the screen shows.
the only enum whose names are known to differ from what the screen shows. The
CorOS 4.1 `Overlord Synth` rename was re-driven across all seven positions on
2026-09-23; it confirms that catalog position 5 displays `WHT` and position 6
displays `PNK`, so the 4.1 enum now carries the same guarded correction as the
4.0.1 snapshot.

### New: `options.OPTION_USAGE` says how many parameters each option list decides

Expand Down Expand Up @@ -317,7 +354,7 @@ on the screen. No library code branches on it.
New `Parameter.display_pos`. A model's parameters come in wire order, and that is
not always the order the unit puts them on screen: a Solo 100 Lead draws `GAIN`,
`BASS`, `MID`, `TREBLE`, `PRESENCE`, `MASTER`, `OUTPUT`, while the wire lists
`MASTER` before `PRESENCE`. Of the 503 models you can place, 338 never carry it;
`MASTER` before `PRESENCE`. On CorOS 4.0.1, of the 503 models you can place, 338 never carry it;
165 carry it somewhere, and on 144 of those the result disagrees with wire order.
If you are showing a block's controls to a person, sort by it and put the
unplaced ones last:
Expand Down
12 changes: 11 additions & 1 deletion docs/STEERING.md
Original file line number Diff line number Diff line change
Expand Up @@ -107,7 +107,7 @@ ADR-0012.
| Keyed grid edits | A mutation is a row/column-keyed `Grid` `UPDATE` | The unit applies grid updates by key and ignores a wholesale preset write ([`architecture.md`](architecture.md), "write_preset is a trap") | `QuadCortex.set_bypass` | Reads, and operations outside the grid |
| One translation boundary | Screen values become wire values in one package, and a source-reading test proves no other module in the package does it | A wrong row is silent: the write lands on a real row and reads back perfectly (design principle 5 in [`domain-model.md`](domain-model.md); ADR-0013) | `pyquadcortex/device/translate/` | The protocol layer keeps zero-based coordinates and quotes the catalog's own units (ADR-0016) |
| Model state goes through the cache | A property reads `Device.state.value(entry, field)`; what it tracks is a `StateEntry` in `device/entries.py` | One account of what the model believes and how it learned it (ADR-0011) | `Device.firmware` in `pyquadcortex/device/device.py` | Values derived from an entry compute from `value()` instead of caching beside it |
| Catalog for structure, a reading for presentation | Option count, wire index and parameter index come from the catalog; anything a person sees needs a reading | Every fixed list the loaded preset reaches lands where the catalog says, re-driven each hardware run. The drawn order of three lists and one list's option names disagreed with the file; `display_pos` has matched twice and rests on those two readings | `tests/hardware/test_option_structure_on_unit.py` | Anything drawn |
| Catalog for structure, a reading for presentation | Option count, wire index and parameter index come from the catalog; anything a person sees needs a reading | Every fixed list the loaded preset reaches lands where the catalog says, re-driven each hardware run. The drawn order of three lists and one list's option names disagreed with the file; `display_pos` has matched twice on 4.0.1 and rests on those two readings. Its catalog populations are profile-specific; see `domain-model.md` | `tests/hardware/test_option_structure_on_unit.py` | Anything drawn |
| Evidence-stamped option lists | Each option list carries a status saying whether a person read its names off the unit | The catalog's `stepNames` differ from the unit's own `dynamic_steps` at 18 of 20 shared positions, and two of the fourteen lists read on the screen did not match it, so an unchecked list must not look checked | `options.OPTION_AUDIT` and `tests/fixtures/catalog/option_readings.json` | A list the unit does not draw is `absent` by observation |
| The profile is the class | `connect()` resolves `(device_type, zenos_git_hash)` to a client class before the handshake; a subclass declares what differs and refuses what it has not verified | One `if firmware ==` in a method body is what polymorphism removes (ADR-0020) | `QuadCortex41` in `pyquadcortex/protocol/profiles.py` | `ALWAYS`: the lifecycle methods every profile needs to connect and clean up |

Expand Down Expand Up @@ -255,6 +255,16 @@ behind each one is in the lab repository,
- **Scope:** `tests/hardware/`, its readme, `tests/test_hardware_report.py`.
No decision record: ADR-0005 already promises this.

### 2026-09-19 - The CorOS 4.1 profile owns its constants snapshot

- **What changed:** `QuadCortex41` binds its contributed models, parameters and
options and declares only operations backed by profile-specific hardware
evidence. Unversioned constant modules remain the 4.0.1 compatibility API.
- **Why:** ADR-0020 makes firmware-specific constants and evidence properties
of the resolved profile; sharing 4.0.1 constants silently mislabels 4.1 wire
values.
- **Scope:** generated 4.1 snapshot, profile registry, tests and public docs.

### 2026-09-17 - The environment is held to the pins it claims to satisfy

- **What changed:** `tests/test_packaging.py` compares each installed version
Expand Down
6 changes: 4 additions & 2 deletions docs/api.md
Original file line number Diff line number Diff line change
Expand Up @@ -222,8 +222,10 @@ qc.remove_block(Block(0, 5))
qc.save_current_preset(Setlist.USER, "30A", "My Patch")
```

`pyquadcortex.protocol.models` has constants for the **414 factory blocks** every
unit has, grouped by category. Anything else, purchased plugin models and your own
`pyquadcortex.protocol.models` is the CorOS 4.0.1 compatibility snapshot and
has constants for its **414 factory blocks**. A connected client's `qc.models`
uses that unit's profile; the contributed CorOS 4.1.0 snapshot exposes 422.
Anything else, purchased plugin models and your own
Neural Captures, has ids that differ per unit, so look those up on the connected
unit through `qc.catalog`:

Expand Down
40 changes: 25 additions & 15 deletions docs/domain-model.md
Original file line number Diff line number Diff line change
Expand Up @@ -1237,15 +1237,19 @@ twice: a cab's four visible controls on 2026-09-11, and a Solo 100 Lead on
`MASTER`, `OUTPUT`; the wire lists `MASTER` before `PRESENCE`). A third reading
that disagreed would unseat it.

| population | count |
|---|---|
| models in the catalog | 533 |
| models a user can place (not hidden, not internal, not in a hidden category) | 503 |
| of those, placing at least one visible control | 163 |
| of those, disagreeing with wire order | 142 |
| placing only some of their visible controls | 23 |
| placing two controls at the same number | 1 |
| counting hidden parameters too: placing / disagreeing | 165 / 144 |
| population | CorOS 4.0.1 | CorOS 4.1.0 |
|---|---:|---:|
| models in the catalog | 533 | 633 |
| models a user can place (not hidden, not internal, not in a hidden category) | 503 | 600 |
| of those, placing at least one visible control | 163 | 234 |
| of those, disagreeing with wire order | 142 | 187 |
| counting hidden parameters too: placing / disagreeing | 165 / 144 | 236 / 189 |

The 4.0.1 figures were measured on 2026-09-15; the 4.1.0 counts were read
from the live ModelRepo on 2026-09-27. The two screen readings above remain
4.0.1 observations and are not re-measurements of the 4.1.0 layout.
On 4.0.1, 23 models place only some visible controls and one places two at the
same number; these two shape counts have not been re-derived for 4.1.0.

So a sort by `display_pos` is not a complete layout, and what the unit does with
an unplaced control is unmeasured. `tests/hardware/test_option_structure_on_unit.py`
Expand All @@ -1254,7 +1258,8 @@ still uses the index.

### `<Padding>` is what a block reserves, and the budget is unknown

A child element rather than an attribute. 331 of 533 models carry one, holding
A child element rather than an attribute. On CorOS 4.0.1, 331 of 533 models
carry one; on CorOS 4.1.0, 398 of 633 do. They hold
`cpu`, `dm_heap`, `pm_heap` and `sw`, and more rarely `dm`, `pm`, `sd_heap`, `nw`,
`dm_hp`. `Model.resources` publishes the numbers under the catalog's own names and
claims nothing more. On 2026-09-15 the loaded preset's free row was filled with a
Expand Down Expand Up @@ -1299,8 +1304,8 @@ hidden, and they depart from the catalog anyway by drawing circles instead of
words. Their model is `internal` and sits in a hidden category. Those are two
more flags, and neither one is `Parameter.hidden`.

Nineteen controls have been read - the fixture holds 24, but five of those are
records of looking and finding no control. Two shorten a word and four draw
Twenty controls have been read - the fixture holds 25, but five of those are
records of looking and finding no control. Three shorten a word and four draw
circles. The other thirteen match the catalog exactly.

Two things stay unknown: whether a control nobody has looked at yet draws its own
Expand Down Expand Up @@ -1339,12 +1344,17 @@ tab called Oscillator, drawn as waveform icons. `OSC1 ACTIVE` sits beside it, on
the same tab, and the flag marks that too. So the flag predicts the screen most
of the time and not always. Whether it predicts writability is untested; see
[section 13](#13-still-open). `options.OPTION_AUDIT` does not use it, and
nothing in the library branches on it. It is also not a boolean: 649 parameters
say `"true"` and one says `"atma"` (the Freeze block's `MOMENTARY`), the Quad
Cortex Mini's `device_type`, so the catalog names the model a parameter is hidden
nothing in the library branches on it. It is also not a boolean: 4.0.1 has 649
parameters saying `"true"` and 4.1.0 has 795; each has one `"atma"` (the Freeze
block's `MOMENTARY`), the Quad Cortex Mini's `device_type`, so the catalog names
the model a parameter is hidden
on. The Soldano carries two parameters called `CHANNEL`, one flagged and one not,
and the screen draws the second only, so the unit honours the flag per parameter.

CorOS 4.1 renames model 30001 to `Overlord Synth`; all seven `OSC1 WAVE`
positions were driven there on 2026-09-23 and produced the same `SIN`, `TRI`,
`SAW`, `SQR`, `PUL`, `WHT`, `PNK` screen sequence.

### `mid_string` is the label at the middle of the wire

Wire 0.5, measured 2026-09-11: a mono cab's `PAN` and a stereo cab's `BALANCE`
Expand Down
Loading
Loading