docs: Exception stack traces on the "Datatypes and Datashapes" page

[chrispy-snps]

Please describe the issue

On the "Datatypes and Datashapes" page, there are some exception stack traces in the example results:

https://ibis-project.org/concepts/datatypes

I did not check other documentation pages for similar exceptions; that might be worth a look, if there's a programmatic way to do it.

Code of Conduct

• I agree to follow this project's Code of Conduct

[gforsyth]

Hey @chrispy-snps -- thanks for raising this!

That exception is included there by design, to show that you cannot arbitrarily combine columns from two separate relations without first joining them.

The exception is so long, that I think we might benefit from not actually executing the failing cell, but sort of "faking" it and only showing the relevant portion of the traceback to improve readability.

[chrispy-snps]

Thanks @gforsyth!

Oops, I did not realize it was included by design - I was quickly scrolling through to look for something else, and the length threw me off. In our product's documentation, I do typically abstract away non-relevant parts of documented traced with ... to focus on the relevant bits of the trace. I wonder if there's a way in your doc framework to automatically post-process traces to the last N levels?

[lostmygithubaccount]

automatically post-processing is probably possible, but for now we can at least just fake this one's output

[cpcloud]

It seems like we should understand why we're showing the exception three times.

We're showing:

1. The rich styled one
2. The IPython one
3. The IPython one again

Not all of our docstrings that show exceptions have this problem, so we should understand what's happening before hacking around it.

[cpcloud]

For example, the Table.pivot_longer docstring shows some intentional exceptions, and the traceback is shown only once (in the IPython style).

[cpcloud]

It's probably because we're calling console.print_exception() and we almost certainly shouldn't be.

[cpcloud]

The second repr is from _repr_mimebundle_.

[chrispy-snps]

Thank you everyone!