This might be a far fetched request, but thought I would write it anyway.
While Jupyter is really great, I think I was most productive when using RStudio. Notebooks are meant to tell a story, step by step. However, most of the time data scientists are just trying different things and exploring the data.
I personally loved that any code I send in RStudio gets executed in the terminal and gives a simple output, and whenever there is a plot it is displayed in a separate panel.
Would it be possible to incorporate an alternative two part output panel - one with a terminal where results are printed and people can write small queries (I know this can sort of be done in the VSCode terminal, but it would require manual connection to the kernel), and a second one where all the images/plots are tunneled into.
There would be no need to show the code that was used, only a history of outputs, exactly the same as in RStudio. Having a variable explorer and help panel would be amazing additions to that, thought of less importance atm. Even Jupyter is trying to get closer to RStudio experience with development JupyterLab.
If this is not possible, perhaps you could consider having an option to:
(1) Run chunks of code without creating a notebook card (new key shortcut, and if you incorporate "# %%" marker, an alternative one for running without a card).
(2) Additional on/off button so that the kernel keeps overwriting/reusing the old card.
This might be a far fetched request, but thought I would write it anyway.
While Jupyter is really great, I think I was most productive when using RStudio. Notebooks are meant to tell a story, step by step. However, most of the time data scientists are just trying different things and exploring the data.
I personally loved that any code I send in RStudio gets executed in the terminal and gives a simple output, and whenever there is a plot it is displayed in a separate panel.
Would it be possible to incorporate an alternative two part output panel - one with a terminal where results are printed and people can write small queries (I know this can sort of be done in the VSCode terminal, but it would require manual connection to the kernel), and a second one where all the images/plots are tunneled into.
There would be no need to show the code that was used, only a history of outputs, exactly the same as in RStudio. Having a variable explorer and help panel would be amazing additions to that, thought of less importance atm. Even Jupyter is trying to get closer to RStudio experience with development JupyterLab.
If this is not possible, perhaps you could consider having an option to: (1) Run chunks of code without creating a notebook card (new key shortcut, and if you incorporate "# %%" marker, an alternative one for running without a card). (2) Additional on/off button so that the kernel keeps overwriting/reusing the old card.
Many thanks for considering this!