| Filename | Latest commit message | Latest commit date |
|---|---|---|
8e4d447d14
These are the only lints clippy 1.99.1 reports that clippy 1.98.0 does not.
Every cargo root in the tree was compared on both toolchains (the evidence is
in the next commit). 1.98.0's clippy also builds every target of the fixed
code, and reports nothing it did not report before, so this commit stands
under either pin.
What changed, and why
- clippy::assert_is_empty is new in 1.99. 36 sites, all in test code
(#[cfg(test)] modules; no lib or bin build reports any):
aac-core and access-core, 23 sites; openpay's op-crypto-extra, op-emv,
op-iso20022, op-reconciliation, op-routing, op-screening, op-settlement
and op-tax, 13 sites. `assert!(x.is_empty())` becomes
`assert_eq!(x, [] as [T; 0])`, `assert_ne!` for the negation, or `""` for
strings, so a failure prints the value instead of only "false". aac-core and
access-core are `#![deny(warnings)]` and `#![deny(clippy::all)]`, so without
this their clippy run fails on 1.99.1.
Judgment call: at 18 of the 36 sites, clippy's own "MachineApplicable"
suggestion does not compile. It names local types by crate-relative path
(`finder::Found`, which is not in scope in a test module), or offers a bare
`[]` whose element type cannot be inferred. So each site names its element
type as its own module sees it (`[] as [Found; 0]`, `[] as [f64; 0]`). The
four lines rustfmt would wrap are wrapped the way rustfmt wraps them.
- clippy::approx_constant (deny-by-default) now knows EULER_GAMMA and
GOLDEN_RATIO. Both are in std::f{32,64}::consts and stable since 1.94; the
jouleclaw workspace MSRV is 1.97.
- jouleclaw-omni, inference/architecture/mochi.rs: `1.618_034` becomes
`std::f32::consts::GOLDEN_RATIO`. They are the same f32
(1.61803400516510009765625), so the latents do not change. This file is
in standards-build.yml's macOS gate (`cargo clippy -p jouleclaw-omni
--features metal -- -D warnings`), which fails on 1.99.1 without the fix.
- jouleclaw-outlier, c_factor: `0.5772156649` becomes
`std::f64::consts::EULER_GAMMA`. Judgment call: the literal was
truncated, so c(n) moves by 3.1e-12. The normalisation in the
isolation-forest paper uses the Euler-Mascheroni constant itself, so this
is the intended value. No other crate and no committed vector reads these
scores, and the crate's tests pass unchanged.
- smart-byte-rs/Cargo.lock: async-trait 0.1.89 becomes 0.1.92, which adds
syn 3.0.9 as its dependency. 0.1.89 put `#[must_use]` on every trait method
it generates. clippy 1.99 now counts the returned `Pin<Box<dyn Future>>` as
must_use already, so clippy::double_must_use fired at 9 #[async_trait] trait
methods. The attribute there is the macro's, not ours, so the right fix is
the macro: 0.1.92 no longer emits it, and every other lockfile in this tree
already uses 0.1.92. A narrow `cargo update -p async-trait --precise 0.1.92`
also moved data-encoding-macro-internal (which accepts syn >=1, <3) from syn
2 to syn 1. That was put back to syn 2.0.117, as on main, so the lock
changes only what the bump needs. `cargo update --workspace --locked` and
scripts/check-lockfiles.sh accept it.
Left as they are
- openpay op-dispute-evidence evidence.rs:92 (clippy::double_must_use) is
reported on 1.98.0 too. Only the message's wording changed in 1.99.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
||
| .. | ||
| conformance/v1 | ||
| jouleclaw-rs | ||
| scales | ||
| vocab | ||
| LICENSE | ||
| LICENSE-SPEC | ||
| README.md | ||
| ROADMAP.md | ||
| SPEC.md | ||
JouleClaw — Energy-Optimized AI Runtime
The substrate for AI that costs Joules, not tokens.
Capability per Joule, not capability per parameter.
What JouleClaw is
JouleClaw is a pure-Rust, omni-modal AI runtime + harness + agentic layer governed by Transaction Science as an open standard (5th in the open-standards quartet, alongside OpenPay, Smart Byte, EOC, and WAI).
Every operation JouleClaw performs walks an explicit energy-tiered cascade, top to bottom, stopping at the first stage that closes the query:
| Tier | Name | Cost class | What runs here |
|---|---|---|---|
| L0 | Cache | picojoules | Content-addressed cache hits |
| L1 | Lawful | nanojoules | Deterministic, pre-compiled primitive (text/code only) |
| L2 | Embed | sub-millijoules | Matryoshka embeddings + nearest-neighbor + hybrid search |
| L3 | Model | joules | Local SSM / ternary / multimodal / diffusion / etc. |
| L4 | Wire | tens of joules | Remote frontier RPC (escape hatch only) |
L1 (Lawful) is only meaningful for text and code modalities. For image, audio, video and 3D generation, the cascade collapses to L0 / L2 / L3 / L4 — the energy savings come from caching plus dispatch to the cheapest model that meets the quality bar.
Why JouleClaw exists
The mainstream "agentic AI" ecosystem treats inference as the entry point. Every state shift, every tool call, every routing decision runs through a 70-billion-parameter stochastic engine, burning kilojoules to do work that deterministic compute or a hot cache could resolve for microjoules.
JouleClaw inverts the default. Inference is the last resort, not the first. Frozen model weights are less trustworthy than current world state; fresh, provenance-stamped retrieval beats synthesis from training cuts. Every operation is metered in microjoules where the hardware permits it, and in millijoules where it doesn't — the spec is honest about the floor on every platform.
The thermodynamic circuit breaker is the safety mechanism. Set an energy budget; when measured consumption exceeds the budget, the breaker trips and the operation halts. Hallucination loops cannot survive an honest joule ledger.
What's here
| Path | Contents |
|---|---|
SPEC.md |
JouleClaw v1 spec — cascade tiers, energy contract, receipt |
CHARTER.md |
Stewardship pattern (shared across open-standards) |
jouleclaw-rs/ |
Apache-2.0 Rust reference implementation |
spec/ |
Tier specifications, energy provenance, receipt format |
conformance/v1/ |
Signed wire vectors for cross-implementation verification |
Reference implementation
jouleclaw-rs/ is the Apache-2.0 reference implementation in Rust. It
runs on Apple Silicon (M3/M4/M5 with UMA + Metal 4 Tensor APIs), AMD Strix
Halo / Ryzen AI Max, Intel Lunar Lake, NVIDIA Jetson, and discrete
NVIDIA / AMD GPUs.
It is omni-modal: text, code, vision, audio, video, image diffusion, 3D Gaussian splatting, and cross-modal fusion all dispatch through the same energy-tiered cascade.
The MCP tool surface is a JCP consumer: jouleclaw-jcp wraps any
jouleclaw-mcp metered tool so every dispatch is decided by the JCP runtime
against a signed, metered Grant before a byte reaches the tool — the tool's
identity is the content-addressed resource, the call's energy estimate is
debited from the grant's joule budget, and every decision (allow and deny
alike) seals as a JCR-1 receipt. An injected instruction can name an ungranted
tool; it cannot make the gate dispatch it.
Status
v0.1.0 reference implementation in active development. The spec is draft status. Cross-implementation conformance vectors will land with v1.0.0.
License
- Code: Apache-2.0 (see
LICENSE) - Spec text: CC-BY-4.0 (see
LICENSE-SPEC)