Skip to content

Comment on linux users of lemmy, how would you fix such idea??

You are viewing one comment and its replies. View all comments · View the comment above

Comments

  • well imagine it as a HTML like

    but designed for app GUI’s on linux

    but instead of being specific to QT or GTK it can work on both and still look normal

    and the actual turning the language into something that GTK or QT understands, is done by a “Translator”, something that turns the code onto something that your DE or Toolkit understands (like a interpreter but instead of machine code its the thing that QT or GTK understand)

    and i am not looking for something like it

    0
    • I’m still not fully understanding you, but if you’re building something, I’d like to suggest that you don’t make it yet another abstraction layer on top of things like QT or GTK. QT and GTK are really heavy and bloated already. Adding something on top will just compound the issue.

      I dunno. I personally don’t care enough about interface/look-and-feel consistency to think it warrants more computational complexity. I’m a big fan of applications that depend directly on things like XCB. I don’t know if there’s an equivalent to XCB for Wayland. I think mostly the Wayland people themselves recommend that people write client applicatons with QT or GTK or whatever, but if I found apps written directly against libwayland, I’d be inclined to prefer those over a QT or GTK alternative.

      1
      • Wayland just gives you a buffer to write to, it doesn’t give you drawing primitives in the way XCB does. It’s kinda the point, and makes implementing display servers/compositors a lot simpler and easier, and most apps will be using some kind of non-XCB toolkit anyways.

        So if you’re making an app you do kinda need a toolkit, but it can be a very simple and lightweight one.

        1
        • most apps will be using some kind of non-XCB toolkit anyways.

          Not my favorites. I’m mostly thinking of the Suckless sort of tools. dmenu, st, dwm, etc.

          But yeah. If a toolkit is necessary for Wayland, then I’d love some extremely minimal options. Or better yet, exactly one minimal option so good that nobody ever feels the need for an alternative. (But that’s not going to happen. Heh.)

          I’ve poked a little bit at the Go GUI toolkit “Fyne”. It appears to use OpenGL/GLFW, which is kinda neat, I think. Looks like GLFW’s dependencies are indeed quite minimal.

          0
    • Knowing some stuff about the APIs of Qt and GTK, it would be a mighty difficult task to do properly. Qt and GTK don’t really “understand” anything, while they both have a declarative markup language (QML for Qt, GtkBuilder XML for Gtk) those are kinda just bolted-on to the imperative/object-oriented cores; that is to say, the abstractions they provide depend on a internal details and external libraries, and it would be a massive pain to design a single language around the constraints and idiosyncrasies of those two.

      So, instead of translating to those existing declarative languages, you’re probably better off just reimplementing your language in terms of the framework-native approach (imperative widget creation and a mainloop for GTK, OOP&signals for Qt). But this is also a lot of work to get right, even if you limit the available functionality.

      An LLM can probably vomit out a prototype for this in an hour, but I feel it will not be very useful to actually make apps.

      In short, you’re probably better off just reading https://wiki.archlinux.org/title/Uniform_look_for_Qt_and_GTK_applications and setting QT_QPA_PLATFORMTHEME=gtk3 or smth.

      0

Sign in to comment. Sign in