Hi there,

Could any kind jupyterlab developer give me an overview of what time-outs 
are important in the jupyter communication protocol?

We have developed a Common Lisp kernel that uses jupyter widgets including 
a few custom jupyterlab extensions (nglview and bqplot).
Our kernel normally takes about 30 seconds to start up because it invokes 
its slow, llvm-based compiler a few times at startup (we are working on 
starting up from images - so this will improve).
This slow startup appears to cause timing problems with the 
jupyter protocol that are very hard to debug.
It seems the jupyter protocol makes assumptions about how fast kernels can 
respond at startup and that might cause problems for other kernel 
developers - it cost me several weeks to figure this out.
The problems go away now that I wrote a fork-server that warms up our 
kernel and then forks children to service notebooks.

I've submitted this as an issue:  
 https://github.com/jupyterlab/jupyterlab/issues/5208


Cheers,

Christian Schafmeister
Professor, Chemistry Department
Temple University


-- 
You received this message because you are subscribed to the Google Groups 
"Project Jupyter" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jupyter/2e596622-5e8b-49e7-888e-edb149faf17d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to