I guess this is some kind of follow-up of #647 , as requested by @thebaptiste
Note: I'm using conda and everything (including conda) from conda-forge
I have tried several times this week to add basemap to a fairly complex (but up-to-date) Python environment, and had to kill conda because nothing happened. That is conda install was not moving anymore, but the conda process was using more and more resources. This time, I killed a 100% CPU conda process using 13 Gb after 200 minutes
So I tried to install only basemap, with conda create -n basemap_test basemap in order to check I could install it at all! This worked perfectly and quickly, but I ended up with numpy 2.3.5 (not numpy 2.4)
If conda was not working with my complex environment, I thought I would use mamba again, which I have not done since mamba became the default solver of conda. And I quickly got a result (no mamba deadlock or guru meditation)!! Trouble is that mamba wants to downgrade the installed numpy, which is a no-go. And also wants to downgrade pyshp (and packaging) but I don't know anything about these
Package Version Build Channel Size
────────────────────────────────────────────────────────────────────────────────
Install:
────────────────────────────────────────────────────────────────────────────────
+ basemap 2.0.0 py312h8d133b9_5 conda-forge 189kB
+ basemap-data 2.0.0 basemap_0 conda-forge Cached
Downgrade:
────────────────────────────────────────────────────────────────────────────────
- numpy 2.5.3 py312he827f4e_0 conda-forge Cached
+ numpy 2.3.5 py312h33ff503_1 conda-forge 9MB
- numpy-typing-compat 20260602.2.5 pyhb860519_1 conda-forge Cached
+ numpy-typing-compat 20260602.2.3 pyhc46ce56_1 conda-forge 13kB
- packaging 26.3 pyhc364b38_0 conda-forge Cached
+ packaging 25.0 pyh29332c3_1 conda-forge Cached
- pyshp 3.1.6 pyhcf101f3_0 conda-forge Cached
+ pyshp 2.3.1 pyhd8ed1ab_1 conda-forge Cached
Summary:
Install: 2 packages
Downgrade: 4 packages
Any thoughts on that, by somebody who knows about packages dependencies and constraints?
Note that the environment where I installed only basemap has pyshp 2.3.1. Could the problem come from pyshp?
Full list of the basemap-only environment below
I have asked the student who wanted basemap to use cartopy instead, so this is not a blocking problem. But it would be nice to still be able to install basemap in a recent environment, if it does not require too much work from the people in charge :-)
(base) $ conda list -n basemap_test
# packages in environment at /home/share/unix_files/cdat/miniconda3_2024-03/envs/basemap_test:
#
# Name Version Build Channel
_openmp_mutex 4.5 20_gnu conda-forge
basemap 2.0.0 py314h9620088_5 conda-forge
basemap-data 2.0.0 basemap_0 conda-forge
brotli 1.2.0 h505cf86_4 conda-forge
brotli-bin 1.2.0 h9908984_4 conda-forge
bzip2 1.0.8 hda65f42_10 conda-forge
c-ares 1.34.8 hebe6cf0_2 conda-forge
ca-certificates 2026.7.22 hbd8a1cb_0 conda-forge
certifi 2026.7.22 pyhd8ed1ab_0 conda-forge
contourpy 1.4.0 py314h5383ef5_1 conda-forge
cycler 0.12.1 pyhcf101f3_2 conda-forge
fonttools 4.66.0 py314hd42368d_0 conda-forge
freetype 2.14.3 ha770c72_2 conda-forge
geos 3.14.1 h55b0958_0 conda-forge
icu 78.3 py310h44b86e0_2 conda-forge
keyutils 1.6.3 h7cc23a3_1 conda-forge
kiwisolver 1.5.1 py314h5383ef5_3 conda-forge
krb5 1.22.2 hbc21106_2 conda-forge
lcms2 2.19.1 h9073bf1_3 conda-forge
ld_impl_linux-64 2.46.1 default_hbd61a6d_102 conda-forge
lerc 4.2.0 hdb68285_0 conda-forge
libblas 3.11.0 11_h4a7cf45_openblas conda-forge
libbrotlicommon 1.2.0 h39a168f_4 conda-forge
libbrotlidec 1.2.0 ha411449_4 conda-forge
libbrotlienc 1.2.0 h018ffa1_4 conda-forge
libcblas 3.11.0 11_h0358290_openblas conda-forge
libcurl 8.22.0 ha042cf0_0 conda-forge
libdeflate 1.25 hd45a770_1 conda-forge
libedit 3.1.20250104 pl5321h373387f_1 conda-forge
libev 4.33 h280c20c_3 conda-forge
libexpat 2.8.4 hd2095e1_0 conda-forge
libffi 3.7.0 h81df57d_1 conda-forge
libfreetype 2.14.3 ha770c72_2 conda-forge
libfreetype6 2.14.3 h5e6c136_2 conda-forge
libgcc 16.2.0 ha9f2e26_7 conda-forge
libgcc-ng 16.2.0 h69a702a_7 conda-forge
libgfortran 16.2.0 h69a702a_7 conda-forge
libgfortran5 16.2.0 h6b99dfc_7 conda-forge
libgomp 16.2.0 he0feb66_7 conda-forge
libjpeg-turbo 3.2.0 hb03c661_1 conda-forge
liblapack 3.11.0 11_h47877c9_openblas conda-forge
liblzma 5.8.3 hb03c661_1 conda-forge
libmpdec 4.0.0 hb03c661_2 conda-forge
libnghttp2 1.68.1 h74cf4be_1 conda-forge
libopenblas 0.3.34 pthreads_hf13c14d_2 conda-forge
libpng 1.6.58 h922cc85_1 conda-forge
libpsl 0.23.1 hd9e3e90_1 conda-forge
libpython 3.14.7 hdc7f604_106_cp314 conda-forge
libsqlite 3.53.4 h13e7031_1 conda-forge
libssh2 1.11.1 h6154650_1 conda-forge
libstdcxx 16.2.0 h934c35e_7 conda-forge
libstdcxx-ng 16.2.0 hdf11a46_7 conda-forge
libtiff 4.7.2 hcc2c06a_1 conda-forge
libuuid 2.42.4 hcfc3c73_0 conda-forge
libwebp-base 1.6.0 hd42ef1d_1 conda-forge
libxcb 1.17.0 hb83e432_2 conda-forge
libzlib 1.3.2 h25fd6f3_3 conda-forge
matplotlib-base 3.10.9 py314h1194b4b_0 conda-forge
munkres 1.1.4 pyhd8ed1ab_1 conda-forge
ncurses 6.6 hdb14827_1 conda-forge
numpy 2.3.5 py314h2b28147_1 conda-forge
openjpeg 2.5.4 heb1ab33_2 conda-forge
openssl 3.6.4 h781a0a9_0 conda-forge
packaging 25.0 pyh29332c3_1 conda-forge
pillow 12.3.0 py314h50bfbbb_4 conda-forge
pip 26.2.1 pyh145f28c_0 conda-forge
proj 9.8.1 he0df7b0_0 conda-forge
pthread-stubs 0.4 h7cc23a3_1004 conda-forge
pyparsing 3.3.3 pyh5ded981_0 conda-forge
pyproj 3.7.2 py314hf1f854a_5 conda-forge
pyshp 2.3.1 pyhd8ed1ab_1 conda-forge
python 3.14.7 hcd007b5_106_cp314 conda-forge
python-dateutil 2.9.0.post0 pyhe01879c_2 conda-forge
python_abi 3.14 9_cp314 conda-forge
qhull 2020.2 h434a139_5 conda-forge
readline 8.3 hd6e31c0_1 conda-forge
six 1.17.0 pyhe01879c_1 conda-forge
sqlite 3.53.4 h9ffa6c4_1 conda-forge
tk 8.6.13 noxft_h1df4ec4_4 conda-forge
tzdata 2026c h151e31d_0 conda-forge
xorg-libxau 1.0.12 h7cc23a3_2 conda-forge
xorg-libxdmcp 1.1.5 h7cc23a3_2 conda-forge
zlib-ng 2.3.3 hce19668_1 conda-forge
zstd 1.5.7 hb78ec9c_7 conda-forge
I guess this is some kind of follow-up of #647 , as requested by @thebaptiste
Note: I'm using
condaand everything (including conda) from conda-forgeI have tried several times this week to add basemap to a fairly complex (but up-to-date) Python environment, and had to kill conda because nothing happened. That is conda install was not moving anymore, but the conda process was using more and more resources. This time, I killed a 100% CPU
condaprocess using 13 Gb after 200 minutesSo I tried to install only basemap, with
conda create -n basemap_test basemapin order to check I could install it at all! This worked perfectly and quickly, but I ended up withnumpy 2.3.5(not numpy 2.4)If
condawas not working with my complex environment, I thought I would usemambaagain, which I have not done since mamba became the default solver of conda. And I quickly got a result (no mamba deadlock or guru meditation)!! Trouble is that mamba wants to downgrade the installed numpy, which is a no-go. And also wants to downgradepyshp(andpackaging) but I don't know anything about theseAny thoughts on that, by somebody who knows about packages dependencies and constraints?
Note that the environment where I installed only basemap has
pyshp 2.3.1. Could the problem come from pyshp?Full list of the basemap-only environment below
I have asked the student who wanted basemap to use cartopy instead, so this is not a blocking problem. But it would be nice to still be able to install basemap in a recent environment, if it does not require too much work from the people in charge :-)