LSP script support jump to definition for sagemaker sdk #19
Conversation
athewsey
left a comment
athewsey
left a comment
There was a problem hiding this comment.
When I tested this, my first Studio user (which also had jupyterlab-s3-browser installed and a couple other minor tweaks, and a workspace with Data Wrangler + one notebook open) ended up in an unresponsive state that I've still not been able to get it out of... Even after removing the LCConfig, trying to delete the .lsp_symlink folder, restarting JupyterServer app, etc.
Haven't yet diagnosed exactly what's up with it or whether this symlinking change was to blame or some other aspect (since I had modified the script to still use S3 browser alongside) - Just a note of caution!
On a different user, everything seems OK (so long as I rm the .lsp_symlink folder as mentioned in comment). However I'm not seeing the jump to source links yet? Maybe I'm just not looking in the right place.
| # Install sagemaker sdk & scikit-learn onto jlab server for source code jump to definition | ||
| pip install sagemaker scikit-learn | ||
| # Create an lsp symlink directory so jupyterlab-lsp can open source-files not found in user home directory | ||
| mkdir .lsp_symlink |
There was a problem hiding this comment.
This fails on 2nd run onwards because the folder already exists.
When I tried adding the -p flag to allow this, I got an error in the line below due to the target already existing and being a (symlink to a) directory:
ln: failed to create symbolic link ‘.lsp_symlink/opt/opt’: Permission denied
So instead I'm currently testing with:
rm -rf .lsp_symlink && mkdir .lsp_symlink
| jlpm add --dev bash-language-server dockerfile-language-server-nodejs | ||
|
|
||
| # Install sagemaker sdk & scikit-learn onto jlab server for source code jump to definition | ||
| pip install sagemaker scikit-learn |
There was a problem hiding this comment.
I guess difficult to know where to draw the line on what libraries to install for this purpose... Would be ideal if jupyterlab-lsp could properly use the kernel environments instead, but agree demoing sagemaker + one third-party could be a nice compromise to show users what's possible
2 participants
Issue #, if available:
Description of changes:
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.