Closed jwodder closed 1 year ago
hm, there is no binary distribution?
I wonder now if we could/should just use their docker container with singularity in it then -- it would be quite an abomination but it works AFAIK, as we use it in https://github.com/ReproNim/containers/blob/master/scripts/Dockerfile.singularity-shim (our "customized" a little image) the quay.io/singularity/singularity:v3.9.0-slim
docker container.
I think it might be more preferable since there is more ways/reasons on why building from source could fail I guess.
@yarikoptic Only the latest versions (in the 3.x series) have prebuilt DEB files available. I had to switch to using them because, even after getting Singularity 2.x to compile, it didn't work.
@yarikoptic Ping.
hm, there are tests fails which are presumably unrelated but never seen before
FAILED ../datalad/tests/test_tests_utils_pytest.py::test_serve_path_via_http[True-None-test1.txt] - ssl.SSLError: [SSL: KRB5_S_TKT_NYV] unexpected eof while reading (_ssl.c:2570)
14213
FAILED ../datalad/tests/test_tests_utils_pytest.py::test_serve_path_via_http[True-None-test_fpath1] - ssl.SSLError: [SSL: KRB5_S_TKT_NYV] unexpected eof while reading (_ssl.c:2570)
14214
FAILED ../datalad/tests/test_tests_utils_pytest.py::test_serve_path_via_http[True-None-test_fpath2] - ssl.SSLError: [SSL: KRB5_S_TKT_NYV] unexpected eof while reading (_ssl.c:2570)
14215
FAILED ../datalad/tests/test_tests_utils_pytest.py::test_serve_path_via_http[True-None-file with space test4] - ssl.SSLError: [SSL: KRB5_S_TKT_NYV] unexpected eof while reading (_ssl.c:2570)
14216
FAILED ../datalad/tests/test_tests_utils_pytest.py::test_serve_path_via_http[True-None-\u0414\u0436\u044d\u0439\u0441\u043e\u043d] - ssl.SSLError: [SSL: KRB5_S_TKT_NYV] unexpected eof while reading (_ssl.c:2570)
14217
FAILED ../datalad/tests/test_tests_utils_pytest.py::test_serve_path_via_http[True-None- |;&%b5{}'"<>\u0394\u0419\u05e7\u0645\u0e57\u3042 .datc ] - ssl.SSLError: [SSL: KRB5_S_TKT_NYV] unexpected eof while reading (_ssl.c:2570)
I really don't see how it could relate to singularity used for build env of git-annex, let's proceed and analyze those fails separately if they do appear on the next cron job
re that KRB -- from https://github.com/dask/distributed/issues/5607#issuecomment-1006029326 seems like some inherent python 3.7 and openssl version gotcha. if persists -- given that 3.7 EOLs soon (27 Jun 2023), we might as well just go for 3.8 at once here since our goal here is not really sweeping python versions, but rather test compatibility with git-annex, so either of those should be as good. Although I am still curious why we get this failure only in this repo/workflows...
Closes #146.