Scrollable region for non-text outputs

[rebornix]

When inspecting data frames, the output is not rendered in a scrollable region, making the whole notebook hard to navigate

[shtratos]

It'd be amazing if we could have a rich and adaptable viewer for dataframes since they are so ubiquitous.
(not as heavyweight as DataWrangler though - it disrupts the context too much and changes VSCode layout).

The very particular pain I'm very frequently running into:

• I get the dataframe in the output of a cell - I want to look at the data to either explore it or to make sure it has the right values and columns.
• the frame is too large to fit, so some columns and rows are elided ...
• now I have to go back and either add df.head() to reduce number of rows or df.columns to see all the columns, or do df.head().T to see all columns in a list with some values.

[shtratos]

Even cooler would be to also have an easy way to reference the dataframes in notebook as context to Copilot Chat for "how do I join/pivot/aggregate/visualise?" types of questions.

[rebornix]

@shtratos thanks for the feedbacks!

Even cooler would be to also have an easy way to reference the dataframes in notebook as context to Copilot Chat for "how do I join/pivot/aggregate/visualise?" types of questions.

Currently we embed the variable summaries in the chat context so you can ask such questions, say you have a data frame df, you can ask how do I join/pivot/aggregate/visualise the data frame and Copilot is able to generate code with the right data frame reference. Let me know if it works as expected for you.

[amunger]

duplicate of #190426