alt-glitch 9fb5e24c9f refactor(platform): machine.facts names the NVIDIA platform from nvidia_platform() ...

The salvaged host summary kept its own Spark detector: win32 + arm64 +
an NVIDIA GPU, or a dgx|spark|gb10 regex on the CPU name. That is one
of the readers the previous commit's products.nvidia_platform() replaces,
and the two disagree (the regex calls any CPU string with "spark" in it a
DGX Spark).

summary() now asks nvidia_platform() once and answers it as
nvidia_platform (rtx_spark | dgx_spark | discrete | null) instead of the
is_spark bool. machine_kind is "Spark" for the two Spark platforms.
machine.facts has no client yet, so the contract field is renamed, not
kept beside the new one.

2026-10-13 22:08:20 +05:30

..

__init__.py

…

test_declaration.py

…

test_facts.py

fix(platform): RTX Spark needs the N1X marker, not just an NVIDIA GPU

2026-10-13 22:07:20 +05:30

test_gpu_adapters.py

fix(local-models): size a Windows integrated GPU from what Windows lets it allocate

2026-10-09 00:44:27 -04:00

test_host_summary.py

refactor(platform): machine.facts names the NVIDIA platform from nvidia_platform()

2026-10-13 22:08:20 +05:30

test_import_hygiene.py

feat(platform): host summary for the desktop questionnaire

2026-10-13 22:07:46 +05:30

test_products.py

fix(platform): RTX Spark needs the N1X marker, not just an NVIDIA GPU

2026-10-13 22:07:20 +05:30

test_resolver_app.py

ci: apply all automatic linter fixes available

2026-10-08 17:22:54 +02:00

test_resolver_core.py

…