Open jacksund opened 2 years ago
As an update, I wrote a simple class based on fabric
to make this process easier for me. The code might be useful if your team decides to implement these features. You can find that code here.
EDIT: switched reference code to an older version that doesn't use a jobqueue
This is a very interesting idea and would simplify the process to connect to remote kernels a lot. Thanks for the suggestion!
We'll try to implement it in a future release.
Any updates?
No, sorry. We've been asking for funding to work on this for the last two years but we haven't gotten it yet.
Now, most IDEs provide a direct and simpler way to connect to a remote server over ssh,
VS code through remote VS code extension
JetBrains through Gateway
Spyder is the only one missing.
3. Spyder is the only one missing
Hello! This is a feature-request, but I hope opening an issue is fine.
EDIT: The original issue has been rewritten for clarity
Current implementation
Users must follow along with the remote kernel tutorial. This involves...
python -m spyder_kernels.console
Problem
The steps above can be very tedious because it involves multiple terminal and file transfers. In most cases, I'll just stick with ipython on the remote cluster, rather than using Spyder.
Possible Solutions
(partial solution) Have
python -m spyder_kernels.console
print out the connection json, which can be copied directly into the Spyder dialog. It'd be an alternative to selecting a file here:From the spyder dialog, pass the SSH connection information to fabric. Then use fabric to run
python -m spyder_kernels.console
and grab the connection info. Spyder then starts a new console using this connection info and the original SSH connection.For solution 1 --> Based on the ipython recipe, it might be possible by having spyder call
%connect_info
as soon as the kernal is started. I know very little about kernels though.