Closed moorepants closed 10 months ago
Here is the difference in the two environments:
$ diff --color /tmp/working.txt /tmp/notworking.txt
21,22c21,22
< cython 3.0.1 py38h17151c0_0 conda-forge
< docutils 0.20.1 py38h578d9bd_0 conda-forge
---
> cython 0.29.36 py38h17151c0_0 conda-forge
> docutils 0.18.1 py38h06a4308_3
28d27
< icu 73.2 h59595ed_0 conda-forge
35c34
< ipopt 3.14.12 he6d3896_1 conda-forge
---
> ipopt 3.14.12 hf9e1ecf_0 conda-forge
53d51
< libhwloc 2.9.1 nocuda_h7313eea_6 conda-forge
60d57
< libspral 2023.08.02 h2baf039_0 conda-forge
67d63
< libxml2 2.11.5 h232c23b_1 conda-forge
72c68
< metis 5.1.0 h58526e2_1006 conda-forge
---
> metis 5.1.1 h59595ed_0 conda-forge
107,110c103,106
< sphinx 7.1.2 pyhd8ed1ab_0 conda-forge
< sphinxcontrib-applehelp 1.0.4 pyhd8ed1ab_0 conda-forge
< sphinxcontrib-devhelp 1.0.2 py_0 conda-forge
< sphinxcontrib-htmlhelp 2.0.1 pyhd8ed1ab_0 conda-forge
---
> sphinx 7.0.0 pyhd8ed1ab_0 conda-forge
> sphinxcontrib-applehelp 1.0.2 pyhd3eb1b0_0
> sphinxcontrib-devhelp 1.0.2 pyhd3eb1b0_0
> sphinxcontrib-htmlhelp 2.0.0 pyhd3eb1b0_0
112,113c108,109
< sphinxcontrib-qthelp 1.0.3 py_0 conda-forge
< sphinxcontrib-serializinghtml 1.1.5 pyhd8ed1ab_2 conda-forge
---
> sphinxcontrib-qthelp 1.0.3 pyhd3eb1b0_0
> sphinxcontrib-serializinghtml 1.1.5 pyhd3eb1b0_0
Something changed in the conda-forge stack and with the libmamba solver it pulls in an ipopt that doesn't segfault.
So the libmamba solver allows Python 3.6 and Python 3.7 builds to solve now, but they both segfault. Suprisingly libmamba finds a set of packages for Python 3.8-3.10 that don't segfault.
Here are the packages that are installed in the working Python 3.8 environment:
and here are the packages installed in the segfault 3.8 from #102 :