Closed branchvincent closed 3 months ago
It looks like that "I" definition is coming from MacOS code, so I'm curious whether that is an issue on other platforms. (If you try to #undef an undefined variable, is that an error?) I think the CI will tell us that.
Thanks for this!
Is circleci running on PRs? I don't see it here https://app.circleci.com/pipelines/github/dstndstn/astrometry.net
But it's from complex.h
, so not just macos. On ubuntu for example:
In file included from /src/.venv/lib/python3.12/site-packages/numpy/_core/include/numpy/npy_common.h:376,
from /src/.venv/lib/python3.12/site-packages/numpy/_core/include/numpy/ndarraytypes.h:4,
from /src/.venv/lib/python3.12/site-packages/numpy/_core/include/numpy/ndarrayobject.h:12,
from /src/.venv/lib/python3.12/site-packages/numpy/_core/include/numpy/arrayobject.h:5,
from pyspherematch.c:17:
pyspherematch.c: In function ‘KdTree_get_data’:
pyspherematch.c:338:17: error: expected identifier or ‘(’ before ‘__extension__’
338 | npy_uint32* I;
Looks like there are some python issues as well https://app.circleci.com/pipelines/github/dstndstn/astrometry.net/256/workflows/89a6156a-855b-40d7-b37f-a842f490f39c/jobs/957
Ah oops, see #299
Fixes #297
Numpy 2.0 removed
eslize
Also, not sure how this is typically handled but
numpy/arrayobject.h
now includescomplex.h
, which defines anI
macro conflicting with local variables