Example notebooks not working

[XouSGC]

Hello,

I was trying the features of the library with the notebooks available at https://resourcecode-project.github.io/py-resourcecode/auto_examples/index.html. The first one about database exploration (plot_0_Database_exploration.ipynb) and the fourth about operation planning (plot_3_Operation_Planning.ipynb) work just fine but the following ones are encountering errors :

• plot_1_data_extraction.ipynb => I get a 500 error coming from urllib when retrieving the 2D_spectrum. Weirdly enough, I can download the file without any problem using a simple wget in the CLI.
• plot_2_Multivariate_Extremes.ipynb => I get an "AttributeError: scipy.stats.mvn has no attribute mvnun; furthermore scipy.stats.mvnis deprecated and will be removed in SciPy 2.0.0". I am using scipy 1.15.2 in my environment. I guess I have to downgrade it for this to work, but this should appear in the library requirements. Or the function could use the new implementation of mvnun which seems to correspond to this (by taking the cdf ).
• plot_4_Producible.ipynb => I get a 500 error coming from urllib when retrieving the 1D_spectrum, same as above.

Is there anything I am doing wrong ? Thanks in advance for your help.

[NRaillard]

Hi @XouSGC ,
Thanks for you interest in the database and your comments. Some hints about the errors you encountered:

• plot_1_data_extraction.ipynb The error may come from a temporary error while reading the IFREMER's FTP, or a limit due to too many requests ?I have switched to the HTTP protocol on the R version of the package, and it seems to be working better;
• plot_2_Multivariate_Extremes.ipynb I agree that I should have pinned the version of scipy to avoid such issues. Downgrading it should make it, but I am not sure about the version. The issue in changing to the new implementation is that it would brake up the package for older version, or we would need to switch depending on the versions. I don't really know how to do, or too little time to work on that...
• plot_4_Producible.ipynb it may be a temporary issue as well.

I think that you are doing well in using the package, but the technical debt is increasing as we have too few people to maintain the toolbox.

[XouSGC]

Hi @NRaillard, thank you for your kind reply. I will keep exploring the different versions of scipy and will post my progress on the thread.

[XouSGC]

Hello again,
As it turns out, the last scipy version that does not produce the error seems to be the 1.13.1. It is not evidently clear from the scipy changelog of the 1.14.0 version what has caused the deprecation : https://github.com/scipy/scipy/releases/tag/v1.14.0
I think the change in question may be in "Expired deprecations" => "Objects that weren't part of the public interface but were accessible through deprecated submodules have been removed."

[XouSGC]

As for the other connection problems, we are still blocked by this 500 error (first request in two days and the error pops up instantly, so it is probably not due to a timeout or a too large number of requests)

[XouSGC]

I just noticed that this PR was created just for the scipy dependency problem. It was merged but apparently not properly released (? I am not very familiar with GitHub pipelines)

[XouSGC]

Ok, I've fetched from the R package the URL for the HTTPS protocol instead of the FTP one and indeed it seems to be working correctly. Thnak you for your suggestions @NRaillard !

[NRaillard]

Ok, my bad: I should publish the new version to pypi to avoid problems.

I am surprised by the 500 error, I cannot reproduce it on my side, the functions are doing well.

I close the issue, do not hesitate to open another one if the problem persist.