aprsa
commented
2 months ago @Raindogjones, @kecnry: can one of you please try to reproduce the segfault on macs and send me the traceback? I can't reproduce it on my end.
Closed aprsa closed 2 months ago
aprsa
commented
2 months ago @Raindogjones, @kecnry: can one of you please try to reproduce the segfault on macs and send me the traceback? I can't reproduce it on my end.
Raindogjones
commented
2 months ago @aprsa Just tried to reproduce with python3.11 on my intel mac and no problems - suspect this might be related to #850 and specific to M1/ARM64 macs...
Raindogjones
commented
2 months ago Paging @bpablo to see if it fails for him on his M1
kecnry
commented
2 months ago The segfaults are occurring outside of this PR as well - or at least on feature-blending (so likely a change to the CI runners themselves, perhaps triggering the same or similar issue as #850).
bpablo
commented
2 months ago I can reproduce. There is at least one test that seg faults: dperdt with the following error: Segmentation fault: 11 /Users/bertpablo/.pyenv/versions/3.11.9/lib/python3.11/multiprocessing/resource_tracker.py:254: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown warnings.warn('resource_tracker: There appear to be %d '
kecnry
commented
2 months ago For this PR, I think we can safely ignore the failures as those are being handled in #850 and don't seem to have been made worse here (so let's keep any additional conversation in #850 and review this as-is).
This PR eliminates duplication in memory consumption between the internal passband structures and ndpolator structures. It also optimizes passband loading/saving. Addresses issue 68.