REPL multiline display

[mflova]

Debug adapter definition and debug configuration

I am using the most up to date version of nvim-dap (see the end of this issue for more info) but it seems the REPL cannot represent multi-line based objects (numpy arrays, pandas dataframes...):

There is a very similar to this issue: #169

However, I would say the root cause to be different. The problem in that issue was the replacement of the \n char via text = text:gsub('\n', ' ') -- Might make sense to change this and preserve newlines? in nvim-dap/lua/dap/ui.lua. But this issue was fixed long time ago as this sustitution is not taking place. That's why I create this other issue.

I am using the most up to date commits from this tool:

Debug adapter version

No response

Steps to Reproduce

1. Create any numpy array
2. Visualize it in REPL

Expected Result

I would expect a numpy array rendered in multiple lines.

Actual Result

I see a numpy array unwrapped along the same line

[mfussenegger]

https://codeberg.org/mfussenegger/nvim-dap/pulls/8 would introduce a .format command for the REPL to control the output format. I'm not sure yet if this is the way to go, but would appreciate some feedback in regards to its usability and if it would solve the problem here.

[mflova]

I am out so I could try in a week or so. So far it looks like the correct solution. I would just need to check if it also works with polars or pandas dataframes.

I am just wondering, what's the need of the .format command? Isnt it possible to do it under the pressence or /n? Or does this erase extra information?

[mfussenegger]

Isnt it possible to do it under the pressence or /n? Or does this erase extra information?

It removes the structured information and there are also use-cases for seeing the raw output as is, including \n without having it interpreted as newline.

The __repr__/__str__ based debugging is something that is very common for Python, but not seen to the same extend in some other languages.

[mflova]

If it is something common for some programming languages, would it make sense to add this as the default behaviour if a flag in the configuration of the debugger is activated? In my typical workflow I would be writting .format every single time I use the debugger