You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Wheels workflow dry run on the release commit at the head of main: run 44380533250, 23 jobs passed, the publish jobs skipped.
GPU lane (poe test-gpu) on a CUDA build: passed on an NVIDIA A40 at the release commit 029ea8e (308 passed, 1 skipped in tests; 31 passed, 1 skipped in native; both skips intended; CUDA 12.8, GCC 14.4, cuFFT and cuBLAS). No AMD device was available for a HIP build.
Tag v0.3.0 pushed on its own (on 029ea8e); wheels run 44399881696.
pypi-native, pypi, pypi-gui approved; uploads complete (solprop-native 0.3.0: five wheels; solprop 0.3.0 and solprop-gui 0.3.0: sdist and wheel each).
published-install and published-install-gui green on five platforms: https://github.com/LupoLab/solprop/actions/runs/44399881696 (the Windows on Arm install was re-run once: its first attempt asked PyPI for 0.3.0 before the index showed it).
Read the Docs: /en/stable/ serves 0.3.0, built from the tagged commit (the v0.3.0 version itself is not activated, as for 0.2.0).
Earlier releases: 0.2.0 is not yanked (maintainer: yanking 0.1.0 was a one-off for the numba backend removal).
Benchmark history rows recorded for the release: seven of nine hardware sets recorded from the tag wheels; two CPU sets and the history pull request to follow.
The tag's wheels run (44399881696) had one failed job, test (windows-arm64): one test, test_a_reducing_run_keeps_no_more_alive_as_it_goes[native], out of 20,986 (the same job passed on the same commit in the dry run). It is a measurement artefact of the test, filed separately; the failed job is being re-run, and the publish jobs wait for it.
Activity
jtravs commentedon Nov 18, 2026
The tag's wheels run (44399881696) had one failed job,
test (windows-arm64): one test,test_a_reducing_run_keeps_no_more_alive_as_it_goes[native], out of 20,986 (the same job passed on the same commit in the dry run). It is a measurement artefact of the test, filed separately; the failed job is being re-run, and the publish jobs wait for it.