Request for testing help

[Radu Serban]

Dear Chrono users,

 

In preparation for the upcoming Chrono release, I would like to ask for your help in testing the PyChrono conda packages. 
We are generating conda packages for Linux, Windows, and MacOS (both Intel and ARM), each targeting Python versions 3.8, 3.9, and 3.10.  Packages corresponding to the latest Chrono code have been generated and uploaded in the last few hours (for technical reasons, the MacOS-apple conda packages will be uploaded manually shortly): https://anaconda.org/projectchrono/pychrono/files

 

I would greatly appreciate it if you could download, install, and test one or more of the available packages (depending on what platforms you use and have available).

Please make sure to install the required dependencies before installing the PyChrono package (see https://anaconda.org/projectchrono/pychrono and https://api.projectchrono.org/pychrono_installation.html).
Once you have the necessary dependencies installed, download the appropriate PyChrono package from the ‘Files’ section and then install it from the local tarball (we have noticed some issues with Anaconda picking the correct default package).

 

Please let us know:

• The OS and Python version you have tested
• Any relevant details on the hardware you used
• Whether you had issues running any of the examples distributed with the package (please make sure to run the demos compatible with your PyChrono install; see the note at https://api.projectchrono.org/tutorial_table_of_content_pychrono.html)
  Please provide details and output for any demos that fail.
• Any suggestions on improving the installation and usage documentation in the links provided above

 

Thank you!
--Radu

 

[luis medina]

Hi Radu,

I hope this can be useful to help to test the PyChrono conda packages:

 

I followed the installation steps 1 to 5, as they are neatly indicated at: https://api.projectchrono.org/pychrono_installation.html

For step 6, I installed the following PyChrono module:

pychrono-7.0.0-py39_2426.tar.bz2

 

Details about OS, Phyton version and hardware:

OS: Windows 10

Python: 3.9.1

CPU: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz   1.99 GHz

RAM:8.00 GB

So far, when I run some  FEA demos I got an issue that shows up when importing pychrono.pardisomkl.  For instance, when I run demo_FEA_beamsIGA.py , I got the following issue:

import pychrono.pardisomkl as mkl

  File "C:\Users\Luis Medina\anaconda3\envs\New_env\lib\site-packages\pychrono\pardisomkl.py", line 10, in <module>

    from . import _pardisomkl

ImportError: cannot import name '_pardisomkl' from 'pychrono' (C:\Users\Luis Medina\anaconda3\envs\New_env\lib\site-packages\pychrono\__init__.py)

_________________________________________________________________________________________________________________

The same occurs for demo_FEA_beams_constr.

I haven’t found other issues when running the other demos ( I'll keep testing).

 

Regards,

Luis

--
You received this message because you are subscribed to the Google Groups "ProjectChrono" group.
To unsubscribe from this group and stop receiving emails from it, send an email to projectchron...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/PH0PR06MB8237546BEC71EAE2635D662CA71F9%40PH0PR06MB8237.namprd06.prod.outlook.com.

[Radu Serban]

Hi Luis,

 

Sorry for the late response, but we’ve been working on and fixing these conda packages since.

 

Thanks for helping with this testing.

 

We finally figured out that issues like the one you reported were due to mismatches between dependencies of PyChrono and of dependencies of dependencies.
In particular, MKL is required by both the Numpy and OCC packages (both of which we list as PyChrono dependencies).  The problem is that pythonocc-core expects a very specific version of MKL (namely 2020), so we had to make sure we build the PyChrono conda package with that version *and* ask users to install the MKL v.2020 conda package.

 

If you get a chance, it’d be great if you could try the latest conda package at https://anaconda.org/projectchrono/pychrono (build 2452).  To make sure there are no artifacts from your previous install, please create a new conda environment (I suggest you stay with Python 3.9) and then install the necessary dependencies and then the PyChrono package in the order specified and using the version specified at the above link.

 

--Radu

[luis medina]

Hi Radu,

 

Thanks for your kind answer, I will try you suggest.

One of the keys – I guess–Chrono is getting more users is the detailed feedback you provide to us. Thanks a lot for that.

Happy 2023 ,and best wishes to all Chrono /PyChrono developers and users.

Regards,

Luis

To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/PH0PR06MB8237948270F7E7FC7FEC4AD3A7ED9%40PH0PR06MB8237.namprd06.prod.outlook.com.

[Radu Serban]

Happy New Year to you Luis as well as to all the Chrono users!