feature: enter insert mode after returning from a nested nvim instance invoked via `:terminal`

[yorickpeterse]

Did you check the documentation?

• I have read all the docs

Is your feature request related to a problem? Please describe.

When starting an nvim instance from a :terminal buffer, upon exiting from that instance the mode is set to normal mode and the user is required to manually enter insert mode again. This is a tad annoying when doing for example git commit followed by git push from a :terminal buffer.

Describe the solution you'd like

The ability to automatically enter insert mode again when returning from a nested nvim instance from a :terminal buffer.

Describe alternatives you've considered

Manually entering insert mode technically works, though it's a tad annoying if you intend to run commands after the nested nvim instance exits.

Additional context

No response

[brianhuster]

You can try this snippet

au BufEnter * if &buftype == 'terminal' | startinsert | endif

[yorickpeterse]

@brianhuster That means always entering insert mode when switching to a terminal buffer, which isn't necessarily what you'd always want (e.g. if you intend to copy some output you now have to exit insert mode first).

[brianhuster]

Unfortunately, it doesn't seem easy to tell if a child Nvim instance was opened by a terminal buffer, and which one. A plugin could use jobstart() to start a Nvim instance while you are in a terminal buffer. Or while a child Nvim is starting up, you could change terminal buffer (because it is asynchronous). So it doesn't seem easy.

[brianhuster]

@yorickpeterse I think this autocmd works too

autocmd TabClosed * if &buftype == 'terminal' | startinsert | endif

Please try it

[yorickpeterse]

@brianhuster That does technically work, but only if the previous tab is indeed the tab that contains the original terminal window and that window is now the active window.

I wonder if the best approach would be for Neovim to expose two autocmd's that get fired when running :terminal commands: one before the execution, and one after. You could then use these to determine if insert mode should be enabled again (e.g. based on the executed command, or just always after finishing a command).

An alternative is that the child nvim process queries the parent for its mode, and if that's insert mode it enables it again before it exists. I'm not sure this is possible though. So basically:

1. Child starts
2. Child checks the mode of the parent's active buffer/window
3. If this is a terminal window, check if the mode is insert mode
4. Just before the child exists, if the parent mode was insert mode send a startinsert command to the parent

[brianhuster]

Will it also need to check if the buffer after the tabpage close is the same as the one right before opening a new tabpage?

[yorickpeterse]

@brianhuster Good point, I think that would indeed be needed.