How to check for backspace key/event in TextInputMultiline callback (CallbackEdit flag)

[inflex]

Version/Branch of Dear ImGui:

1.91.9

Back-ends:

SDL2 software

Compiler, OS:

linux + gcc

Full config/build information:

No response

Details:

I'm working on a simple wordwrap for InputTextMultiline for use with monospace fonts and so far "most" things are going well except for when I start backspacing across prior linebreaks ( this is due to the codec being used creating \r\n instances to represent hard-breaks, so backspacing then produces a sole \r ).

My question is, where/how can I capture the backspace keypress and is it available within the callback so that I can properly test for and handle this corner case?

While I can see the callback being invoked during with the backspace press if I have the ImGuiInputTextFlags_CallbackEdit, I however cannot see where I can detect if it was the backspace key that induced that, neither data->EventKey or data->EventChar indicate anything for me.

Screenshots/Video:

No response

Minimal, Complete and Verifiable Example code:

No response

[ocornut]

We cannot provide the info in EventKey for ImGuiInputTextFlags_CallbackEdit as multiple edits may have happened.

You could technically simply poll the key on that event, but the edit will have been done already, and it feels a bit like a quagmire to think you can reliably reconstruct a sequence of event and patch a text buffer this way.

Generally speaking, I'm not super comfortable moving forward with this that as:

• word-wrapping is not yet supported and I don't think pre-wrapping the string is a good solution, proper support would be better.
• ideally I would phase out as many inputtext callbacks in favor of using input routing check + making it easier to externally access and edit input text contents.

So adding another layer of complexity into existing callbacks in order to facilitate developing a workaround for a feature seems like a patch over a patch. But if you can think of something simple I can do I'll try to help.

[inflex]

I appreciate the candid response. I agree a proper WW solution within InputTextMultiline would be the most ideal solution though I know you've got your plate more than full with everything else you're handling, so I'm taking this path as a "bodge" solution until such time that it arrives in native form.

I'll ponder on my options a little more and report back. Thank you again for your time.