xpresspipe won't pass the final installation check xpresspipe test

[hollyjaro]

Hello,

I tried to install the xpresspipe according to the installation tutorial. I did not encounter any warnings or bugs during installation but at the last step to check if everything is ok I am getting following:

Traceback (most recent call last): File "/nethome/hollyj2/miniconda3/envs/expresspipe-env/lib/python3.7/site-packages/numpy/core/function_base.py", line 117, in linspace num = operator.index(num) TypeError: 'float' object cannot be interpreted as an integer

I tried to troubleshoot it by removing conda following the tutorial in FAQ section and do clean install with miniconda3 but I got the same result. I would like to install and try to run the xpresspipe "locally" on our HPC. Please see my command line output attached xpresspipe_float_error.txt

I would appreciate any hints how to solve it. Thanks!

Jaro

[j-berg]

Hi Jaro,

Thanks for your interest in using XPRESSyourself, and thanks for all the details. From what I can tell, my suspicion is that this issue is arising in numpy, but maybe we can dig into this a little more. When I run xpresspipe test, I seem to be able to run everything smoothly, but could you activate your conda environment and send me what you get when you run conda list? Also, if your HPC environment SLURM, or something else. Sometimes, I would run into issues where default packages on the HPC would mess with the conda installations, so I would run modules purge to clear their installations from my environment.

This is the current version of numpy I am using that doesn't seem to raise any issues. The issue could also be related to Seaborn, so I've included my current version of that as well:

python                    3.7.4                h265db76_1  
conda                     4.7.12                   py37_0  
numpy                     1.17.2           py37hf9bf1b6_0  
numpy-base                1.17.2           py37h2f8d375_0  
seaborn                   0.9.0                    py37_0  

[hollyjaro]

Hi, thanks for the fast reply. I tried to purge the preloaded modules first and it didn't help. I checked the versions of numpy and seaborn. The seaborn was the same but the expresspipe installed numpy 1.18.1. After downgrade to 1.17.1 the problem was solved. Now I get the summary after I run the xpresspipe test command. It would help to restrict the installation to that version of numpy by default.

Bug solved!

Thank you! Jaro

[j-berg]

Agreed, looks like 1.18 is new as of this month, so maybe there is some compatibility weirdness with the new version for whatever reason. I will restrict numpy to <=1.17 in the next release (probably this weekend or earlier) and if numpy>=1.18 works at some point in the future will remove the version restriction.

Thanks!