Reuse cached results

[DanielHermosilla]

Hey,

I am new to Quarto, so I am sorry if this has been asked before.

Currently I have some code chunks to execute. My problem goes with one specific code chunk that is computational heavy. I had ran the chunk with Molten and already got the results. However, if I would like to display it with Quarto, it needs to be computed again. Worst case is when it made all of the computation and the preview fails due to a syntax error on another part of the document.

Is there a workaround for this? Maybe this is an expected behaviour that is global among every .qmd.

Thanks

[jmbuhr]

This is more of a general Quarto question rather than Neovim specific, but maybe I can help. Firstly, here are the docs for Quarto on caching: https://quarto.org/docs/computations/caching.html

But this doesn't really solve your question, because caching for python code via jupyter cache is an all or nothing type of thing. If just one code chunk changes, all cells will be recomputed (for R it is different, there each cell has its own cache).

However, three things will be interesting to you:

1. If your document is an ipynb file, not a plain text qmd/quarto file, Quarto will not re-rexecute the code by default and use the output recorded in the ipynb file
2. You can use https://github.com/GCBallesteros/jupytext.nvim to automatically open an ipynb file as a qmd file, converting back and forth
3. https://github.com/benlubas/molten-nvim has a MoltenExportOutput command to export the cell output to a ipynb file.

But here are my two cents about this: If you have a computationally heavy code cell, there is a very high likelihood that this code shouldn't be in a notebook in the first place. There are almost always better alternatives to keep track of results and executions than notebook caches that encourage more reproducible practices. For R this would be the targets package (https://books.ropensci.org/targets/), while for Python there are quite frankly too many, but my current favorite is Hamilton (https://hamilton.dagworks.io/en/latest/). Both of these make it easy to use a notebook as a driver and sort of command center, while the computationally heavy code lives in modules and can (optionally) be cached transparently.

[DanielHermosilla]

This is more of a general Quarto question rather than Neovim specific, but maybe I can help. Firstly, here are the docs for Quarto on caching: https://quarto.org/docs/computations/caching.html

But this doesn't really solve your question, because caching for python code via jupyter cache is an all or nothing type of thing. If just one code chunk changes, all cells will be recomputed (for R it is different, there each cell has its own cache).

However, three things will be interesting to you:

1. If your document is an ipynb file, not a plain text qmd/quarto file, Quarto will not re-rexecute the code by default and use the output recorded in the ipynb file

2. You can use https://github.com/GCBallesteros/jupytext.nvim to automatically open an ipynb file as a qmd file, converting back and forth

3. https://github.com/benlubas/molten-nvim has a MoltenExportOutput command to export the cell output to a ipynb file.

But here are my two cents about this: If you have a computationally heavy code cell, there is a very high likelihood that this code shouldn't be in a notebook in the first place. There are almost always better alternatives to keep track of results and executions than notebook caches that encourage more reproducible practices. For R this would be the targets package (https://books.ropensci.org/targets/), while for Python there are quite frankly too many, but my current favorite is Hamilton (https://hamilton.dagworks.io/en/latest/). Both of these make it easy to use a notebook as a driver and sort of command center, while the computationally heavy code lives in modules and can (optionally) be cached transparently.

Thanks a lot for your response! Your help is much appreciated!