Fork or Revert Kernel?

[Adam Rule]

I am modifying Min's Scratchpad extension and am wondering if there is a way to make it so code executed in the scratchpad does not impact the main notebook. As it is, if I mess with parameters in the scratchpad, it will change those parameters in the notebook as both share the same kernel. Would it be possible to either:

1. Fork the kernel to start a new one for risk-free experimenting (this seems to break the one kernel per notebook paradigm)
2. Revert the kernel to a prior state when the scratchpad is closed?

I'll look through the kernel management source code in the meantime.

[Brian Granger]

JupyterLab has a "Code Console" that is similar to this Scratchpad,
but allows more flexibility in terms of which kernel is used.

As far as 1) and 2) are concerned, I don't know of any general way of
forking, rewinding a full runtime without full blown process level
checkpointing. I am guessing that would take you down a rabbit hole.
You could begin to explore some of the immutable namespace ideas from
my talk at UCSD though. But keep in mind, there a ton of subtleties in
getting that to work robustly (have to serialize state to disk to
avoid growing memory, not all objects are easy to serialize).

> --
> 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 jupyter+u...@googlegroups.com.
> To post to this group, send email to jup...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/jupyter/7bc20957-5fc2-4442-97f2-aee3200d1660%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



--
Brian E. Granger
Associate Professor of Physics and Data Science
Cal Poly State University, San Luis Obispo
@ellisonbg on Twitter and GitHub
bgra...@calpoly.edu and elli...@gmail.com

[Robert Schroll]

This is tangentially related at best, but several years ago, I worked on a project called Reinteract [1] that tried to do something similar.  Working at the AST level, we were able to identify 99% of the mutations that would occur, allowing Reinteract to save and rewind to checkpoints.  Unfortunately, that last 1% proved to be essentially impossible to detect with this approach.

I'm happy to discuss this further if there's interest, but I'll hold off on further thread hijacking for now.

Robert

[1] http://www.reinteract.org/

-- br />Brian E. Granger

Associate Professor of Physics and Data Science
Cal Poly State University, San Luis Obispo
@ellisonbg on Twitter and GitHub
bgra...@calpoly.edu and elli...@gmail.com

-- br />You received this message because you are subscribed too the Google Groups "Project Jupyter" group.

To unsubscribe from this group and stop receiving emails from it, send an email to jupyter+u...@googlegroups.com.
To post to this group, send email to jup...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/jupyter/CAH4pYpTAJuVamp4zaU2rcYDDZiZ-NVGMZLLZu2pCRFBPCsJwUQ%40mail.gmail.com.

[Brian Granger]

Cool, thanks for posting Robert!

> https://groups.google.com/d/msgid/jupyter/local-bc6edb9e-9f58%40mando.

>
> For more options, visit https://groups.google.com/d/optout.



--