Non-Latin alphabet support

[ystyle]

Unable to activate fcitx5 input method

[gabm]

thank you for this. Can you provide more information in what this is and what is useful for?

if you are a developer: do you have resources that demonstrate it?

[ystyle]

Add multiple input methods to fcitx5, use ctrl + space to switch input methods when annotating, but the input method does not respond.
I have never learned rust. Neovide solved this problem in x11 not long ago.

[lilydjwg]

If you can't put a GtkEntry there, you'll need to interact using GtkIMContext yourself.

Update: for why this is useful: An input method is used to input complex scripts like Chinese, Japanese or the like.

Satty doesn't even support paste, which means the usual workaround for lacking IM support doesn't work either. ydotool emulates a keyboard and thus can't type Chinese either. wtype should work, but it has other problems.

[gabm]

i really didn't get what you wanted to do as the interface is mainly pointer based - but you do mean the text box input, right?

[lilydjwg]

Yes. That's the only place where the user can input text, I suppose?

Basically, you need to create a GtkIMContext, call gtk_im_context_focus_in when the user is going to input text, and call gtk_im_context_set_cursor_location when the text cursor moves, listen to commit signal to get the inputted text and insert it to your text buffer. You also need to handle those preedit-{start,changed,end} signals and use gtk_im_context_get_preedit_string to get the preedit string, show it in your UI.

I'll make a screen record to show how text is inputted with fcitx5 in a normal GTK application.

[lilydjwg]

This is how an input method is in use. The purple window is from the input method, and the application window is gtk4-demo-application.

out.mp4

You need to tell the input method that the user is going to input text, where the cursor is, and retrieve and draw both preedit and committed text on your UI.

Although only the text tool needs this, it's an important annotation tool for screenshots, so I really hope it works for us.

[gabm]

I made progress on this. Would you be able to build from source and test?

[ystyle]

I made progress on this. Would you be able to build from source and test?

I can try, but it will have to be tomorrow (it's night here now)

[gabm]

I'll let you know when i pushed it.. still some refactoring needed

[lilydjwg]

I can test too.

[gabm]

Hi both, thank you for volunteering. I pushed a branch multi-input that uses the IMMulticontext as seen in https://github.com/flxzt/rnote/blob/main/crates/rnote-ui/src/canvas/mod.rs#L473. There is currently no "pre-edit" text handling because I don't know how to draw that currently...

Can you let me know if that works and maybe create a screen recording?

[lilydjwg]

No, it doesn't work. No text input (in the wayland protocol) is created. It seems that you created an IMMulticontext but never used it? I don't quite understand what that #[relm4::component] macro is doing though.

[gabm]

sorry to hear that.. how can i test this myself?

[lilydjwg]

Try to install fcitx5? You need to add and activate one input method (one for any language will do).

[Clareses]

Hello, I've submitted a PR for this issue. Please take a look. Thanks. @gabm #52