Bus error when using openneuro-cli on UPenn's cluster

[tsalo]

What went wrong?

I regularly get a Bus error when using the OpenNeuro CLI on UPenn's cluster. I don't know if this is a problem with UPenn's cluster or with openneuro-cli, but I figured I should open an issue.

Expected behavior

When running openneuro login, I would expect it to run successfully. However, after the three prompts, it stalls and then raises a Bus error.

How to reproduce

1. Create a conda environment:

   conda create --name openneuroenv python=3.11
   conda activate openneuroenv
   conda install -c conda-forge nodejs
   conda update -c conda-forge nodejs  # increases it to v20.9.0
   npm install -g @openneuro/cli
   

   • nodejs is v20.9.0
   • npm is 10.1.0
   • @openneuro/cli is 4.22.0
   • bids-validator is 1.14.3

2. Run openneuro login. I chose https://openneuro.org, provided my API key, and chose Yes to error reporting.

Desktop

• Operating System: CentOS Linux 7 (Core)
• Kernel: Linux 3.10.0-1160.21.1.el7.x86_64
• Architecture: x86-64

Additional information

No response

[tsalo]

@effigies proposed creating a Python-based credential manager, which would probably address this problem.

[tsalo]

Can users just use keyring, or does OpenNeuro have specific requirements? Also, is there any way for users to login with the generic credentials? I.e., a username and API key?

[nellh]

openneuro login creates a .openneuro JSON file in the user's home directory and the API key and URL in that file are the only required values.

[tsalo]

@nellh do you know how that JSON file could be used for authentication without using openneuro-cli as the credential helper (e.g., with git-credential-datalad)?