conda\mamba environment file is missing

[lsuchoy]

I'm having issues creating the conda/mamba environment locally to run uw3. The file in the main folder called environment.yml is a symlink to a non-existing file .binder/environment.yml which I assume has to do with the binder repository. Is there an alternative .yml file I can use to create the environment?

Lior

[lmoresi]

Yikes - that’s my bad ! I’ll update this with the real file.

…

Sent from my iPhone On 6 Dec 2024, at 13:10, Lior Suchoy ***@***.***> wrote:  I'm having issues creating the conda/mamba environment locally to run uw3. The file in the main folder called environment.yml is a symlink to a non-existing file .binder/environment.yml which I assume has to do with the binder repository. Is there an alternative .yml file I can use to create the environment? Lior — Reply to this email directly, view it on GitHub<#274>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADABPIZ7ZCDTHAQGW4CZGOT2EGO4XAVCNFSM6AAAAABTEUJ6Q6VHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZXGYZDGMJUGY>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[lsuchoy]

Thanks! The file in the development branch is still a symlink. I manually copied the content of the new file to my local file but the development branch probably needs an update too.

[lmoresi]

Yep - I’ll do that with a PR !

…

Sent from my iPhone On 7 Dec 2024, at 12:21, Lior Suchoy ***@***.***> wrote:  Thanks! The file in the development branch is still a symlink. I manually copied the content of the new file to my local file but the development branch probably needs an update too. — Reply to this email directly, view it on GitHub<#274 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADABPI6N7XKSXI5I5AVMY732ELR2XAVCNFSM6AAAAABTEUJ6Q6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCNBZGM3TOMQ>. You are receiving this because you commented.Message ID: ***@***.***>

[lsuchoy]

Hi, I think there may be something missing in the environment or maybe a step missing in the tutorial. I managed to set up the environment with mamba (with conda it didn't work...) but I keep getting the following error when running test_0004_pointwise_fns.py:
ImportError: cannot import name 'petsc_discretisation' from 'underworld3.cython' (/DATA/underworld3/src/underworld3/cython/__init__.py)
I assume this is because there is either a package missing or the cyton code isn't pre-compiled. If I do need to recompile it, which compiler should I use? the intel c++?

Also, you may want to add to the manual that PATH/underworld3/src/ should be added to PYTHONPATH before testing.

[lsuchoy]

UPDATE: NOT Solved this by adding

import pyximport
pyximport.install()

before
from underworld3.utilities._jitextension
However, this requires running the code under sudo (sudo python test_0004_pointwise_fns.py) so there must be a better solution.

UPDATE: Using sudo fixes an issue in writing data during pyximport.install() but it essentially overrides the mamba environment so it requires having all the modules installed on the base environment, which isn't realistic.
I'm looking for a better solution

[lmoresi]

@julesghub - could you respond on this issue ?

[julesghub]

@lsuchoy Can you point me to the exact environment.yml you used and the commands to load it up. I'll try recreate and fix.

[julesghub]

ok, I have made changes to the development branch to address the conda installation issues.
See e26b413

@lmoresi & @lsuchoy - you MUST pip install with no-build-isolation now. i.e.
pip install . --no-build-isolation

I'll clean this up and then make changes to the install guide.
@lsuchoy thanks for bring this to our attention. Please test this on the development branch of the code and let me know the result.

[lmoresi]

Sounds good. If it works for you, @lsuchoy, let us know and we'll bring this change into main ASAP.

L

[lsuchoy]

Thanks @julesghub for looking into this. I removed the old environment and files and rebuilt everything and the problem seems to be resolved.

One note is that you should probably add python setup.py build_ext --inplace to the section Did it work? right before source test.sh (or just add it to the bash script).

[lsuchoy]

Just to note there were a few tests that did not pass. I'm posting them here:

============== test session starts ============== 
platform linux -- Python 3.11.11, pytest-8.3.4, pluggy-1.5.0
rootdir: /DATA/underworld3/tests
configfile: pytest.ini
plugins: anyio-4.7.0, typeguard-4.4.1
collected 0 items

============== no tests ran in 8.23s ==============
WARNING! There are options you set that were not used!
WARNING! could be spelling mistake, etc!
There is one unused database option. It is:
Option left: name:--config-file=tests/pytest.ini value: tests/test_0050_utils.py source: command line
============== test session starts ============== 
platform linux -- Python 3.11.11, pytest-8.3.4, pluggy-1.5.0
rootdir: /DATA/underworld3/tests
configfile: pytest.ini
plugins: anyio-4.7.0, typeguard-4.4.1
collected 61 items

tests/test_0101_kdtree.py ........                                                                                                                                                                                                    [ 13%]
tests/test_0501_integrals.py .........x                                                                                                                                                                                               [ 29%]
tests/test_0503_evaluate.py ............                                                                                                                                                                                              [ 49%]
tests/test_0504_projections.py ...                                                                                                                                                                                                    [ 54%]
tests/test_0601_mesh_vector_calc.py .........X..................                                                                                                                                                                      [100%]

============== XPASSES ==============
==============  short test summary info ==============
XFAIL tests/test_0501_integrals.py::test_integrate_swarmvar_deriv_00
XPASS tests/test_0601_mesh_vector_calc.py::test_no_ijk
============== 59 passed, 1 xfailed, 1 xpassed in 74.82s (0:01:14) ============== 
WARNING! There are options you set that were not used!
WARNING! could be spelling mistake, etc!
There is one unused database option. It is:
Option left: name:--config-file=tests/pytest.ini value: tests/test_0101_kdtree.py source: command line

[lmoresi]

xfail is deliberate, not a bug
xpass looks like something that is outdated and we should look at