Cell caching about nbclient

[davidbrochart]

It uses the ast module, and it assumes that an output is a variable that is assigned to, and an input is any variable that is accessed.

[davidbrochart]

This is something that can be solved at the kernel level, and akernel tries to support that.
Is it what you have in mind?

[JanPalasek]

I think so. Im curious how input inference is reliable. Seems like a task that would need a python expression parser.

[JanPalasek]

@davidbrochart That's awesome! Do you think it would be possible to integrate it to ipykernel directly? So everyone could benefit from this functionality? Imho a lot of Jupyter projects would benefit from that.

[davidbrochart]

That's awesome!

Thanks!

Do you think it would be possible to integrate it to ipykernel directly?

It's not that easy, ipykernel is the main Python kernel and should be very stable. But it is worth trying, do you want to open an issue there and start a discussion about it?

[JanPalasek]

That's awesome!

Thanks!

Do you think it would be possible to integrate it to ipykernel directly?

It's not that easy, ipykernel is the main Python kernel and should be very stable. But it is worth trying, do you want to open an issue there and start a discussion about it?

Ye, that would be great. If that fails, there might be a backup plan for caching on a bit higher level. I suggested something here: executablebooks/jupyter-cache#89 . Tho it's probably not as good as your implementation since it's higher above the kernel.