docs: update all READMEs, PYPI docs, RELEASE_NOTES, and roadmap to reflect v0.6.2 without macOS Intel x86_64 wheels (due to GitHub Actions queue delays). Supported: Linux, Windows, macOS arm64 + sdist.

Author: qectorlab

PYPI_README.md

@@ -23,9 +23,11 @@ Supported package target for the public release workflow:
 - Python 3.9 to 3.13
 - Linux x86_64 wheels (manylinux)
 - Windows x64 wheels
-- macOS Intel x86_64 + Apple Silicon arm64 wheels
+- macOS Apple Silicon arm64 wheels
 - Source distribution (sdist) for custom/source builds
 
+macOS Intel (x86_64) wheels are not included in v0.6.2 due to extremely long GitHub Actions runner queues (they will be added in a follow-up).
+
 Optional research and validation extras:
 
 ```bash

README.md

@@ -26,7 +26,7 @@ PyMatching-compatible MWPM validation - Belief-matching accuracy mode - BP-OSD f
 pip install qector-decoder-v3
 ```
 
-Supported public wheel target: **Python 3.9 to 3.13** on Linux (manylinux), Windows (x64), and macOS (Intel x86_64 + Apple Silicon arm64) where wheels are published for the release.
+Supported public wheel target: **Python 3.9 to 3.13** on Linux (manylinux), Windows (x64), and macOS (Apple Silicon arm64). macOS Intel (x86_64) wheels will be added in a follow-up release (GitHub Actions Intel runner queues were extremely long). Source distribution (sdist) is available for custom builds.
 
 ### Optional Python extras
 

RELEASE_NOTES.md

@@ -12,6 +12,7 @@
 - Reduced namespace leakage in `__init__.py` (os/sys/subprocess/np no longer pollute the public module).
 - `recommend_decoder` / routing safely avoids recommending UF family on hypergraphs.
 - sdist now published alongside wheels.
+- Wheel matrix for Linux, Windows, macOS arm64 (macOS Intel x86_64 excluded from this release due to 8+ hour GitHub Actions queue times on macos-13; will be added later).
 - Expanded test matrix and relaxed d=21 latency threshold for stability.
 - All docs, versioning, and metadata aligned to 0.6.2.
 

docs/PLATFORM_ARTIFACT_ROADMAP.md

@@ -14,7 +14,7 @@ This roadmap turns the external audit recommendations into explicit public work
 | GPU bit-identity | Tested configurations and docs | Safe as bit-identity claim, not throughput claim |
 | GPU throughput | Not yet a stable public proof asset | Do not claim speedup without regenerated local artifacts |
 | Linux/macOS benchmark corpus | Not yet published as comparable artifacts | Do not imply cross-platform benchmark equivalence |
-| Prebuilt wheels | Published for v0.6.2+ (Linux, Windows, macOS Intel+ARM) | Full matrix in CI; attach attestations |
+| Prebuilt wheels | Published for v0.6.2 (Linux, Windows, macOS arm64). Intel x86_64 excluded due to CI queue delays. | Full matrix in CI (Intel to be added) |
 
 ## Planned artifact classes
 
@@ -77,7 +77,9 @@ As of v0.6.2, CPU-safe wheels are published via CI for:
 
 - Linux x86_64 (manylinux)
 - Windows x64
-- macOS Intel x86_64 + Apple Silicon arm64
+- macOS Apple Silicon arm64
+
+macOS Intel (x86_64) wheels were excluded from this release due to extremely long GitHub Actions runner queues. They will be added in a follow-up.
 
 sdist is also published.
 
@@ -86,7 +88,7 @@ Candidate / supported targets (v0.6.2+):
 ```text
 Linux x86_64 CPython 3.9–3.13
 Windows x64 CPython 3.9–3.13
-macOS x86_64 CPython 3.9–3.13
+macOS x86_64 (Intel) - planned for follow-up (queues too long in v0.6.2)
 macOS arm64 CPython 3.9–3.13
 ```
 
@@ -145,6 +147,6 @@ Unsafe:
 
 ```text
 QECTOR already has universal cross-platform benchmark proof.
-QECTOR ships production wheels for Linux, Windows, macOS Intel x86_64 and Apple Silicon.
+QECTOR ships production wheels for Linux, Windows, macOS Apple Silicon (Intel x86_64 to follow).
 QECTOR GPU throughput is generally faster than PyMatching.
 ```