Snappy Widgets


Widgets that snap into apps.


Today, every time a developer develops an app, it is them who decide how the app should look and feel, not you, the user. However, if you look around, most apps pretty much share the same kind of widgets:

  • all chat apps have conversations list, the input field for text, voice, emojis, files into a chosen conversation, and the view of the conversation itself.
  • all CAD design apps have the object 3D views, tool selectors, and the object/operation trees.
  • all the mail apps have folders and message title lists.

The examples can continue, but the point is, that these widgets repeat across different applications.

The creators of the first operating systems graphical user interface, such as Windows, came up with the paradigm, where developers create a window, and a world within that "window". They had not envisaged that the user interface lowest common denominator should be a "widget", and the user would compose widgets into what they need. So, let's re-imagine a life of a user with "Widgets OS" rather than "Windows OS".

  • Upon starting the snappy widgets environment, a user could instantiate new empty docking nodes to which to dock new components, and start placing widgets into them from a library of possible widgets.
  • Each node with an arrangement of widgets could be saved as an "app".
  • Having such an arrangement, a user could configure the widgets, by looking up APIs available as backends to them, as the sources for these widgets would be configurable, based on the data protocol defined by each widget.
    • For example, if it is a chat entry field widget, perhaps there are many apps with a public API that snaps to this widget.
    • A user would choose, which backends to make available for the specific widget within the docking node of choice.
      • For example, to emulate Trillian chat integrator, a user could choose all supported backend apps for chat entry field, which should auto-toggle based on the active chat window.

Having such a widgets system, front-end developers would not have to repeat their work: new widgets would need to be created only if there is no needed widget in existence. The development and publishing of each widget would have to be a bit like specifying a protocol of data, because the sub-types that the widget supports would be specific. Popular widgets would have many backends providers, and someone who wants to create a new app would not have to create front-end, because the front-end would be composed by user, not the developer.

What's more, we can imagine, that under such system, widgets could be dragged-and-dropped from one docking node to another. It would feel like dragging-and-dropping widgets from one app window to another app window.

The greatest benefit would probably be in the ability to quickly create new apps just by writing appropriate APIs, and without any front-end coding at all, meaning, more developer work would be available for systems engineering and data management, rather than front-end work.

Finally, when it comes to real implementation of this idea, these days there are many widget libraries written for developers (only in web development: Material, Bootstrap, Ant Design, BalmUI, Buefy, Chakra UI, Element, Framevuerk, Oruga, Tachyons, Tailwind CSS, Windi CSS, Vant, Vuesax, etc.), so lots of people are working on them. It would be quite straightforward to take some of the good ones, and make them snappy like legos to a docking point, via introducing a specialized browser extension, an Electron application, and an Android/iOS application, that packages such app-making environment.

However, the web-technologies based UIs tend to be slow, and the non-native apps in general seem to be slower, the next step could be to develop a completely native implementations by desktop environment developers, or perhaps integrated ("Write once. Deploy everywhere.") framework creators, like Let's see.

Inspired by Structured Internet Client.

(别通知) (可选) 请,登录

在我的附加 GUI 概念中,我将 GUI 视为数据流系统。



In my additive GUIs concept I treat the GUI as a data flow system.

In an email client you have a list of folders. Clicking a folder flows emails of that folder to a email list. Clicking an email in the list flows the email into an email preview.

Perhaps widgets can be described by the relationships towards eachother and the data they flow between one another.


让我们在 oo 上实现这个

Let's implement this on oo

    :  -- 
    : Mindey
    :  -- 


// Visual Basic [...]


// Visual Basic [...]

It was for programmers, not for end users. What this idea is about, is a tool for end users, obsoleting app UIs, so in the future apps don't need their own UIs.

Visual Basic在此方面处于业界领先地位,并且功能非常强大。虽然我从未学过如何编程大多数小部件。只是桌子。

Visual Basic was ahead of the industry on this account and was very powerful. I never learned how to program most widgets though. Just tables.




-过于精细(最终用户通常没有时间从头开始构建菜单,或者从头开始构建聊天窗口等) -不是协议定义的(复杂的窗口小部件通常定义了复合数据类型,甚至是交互协议本身,例如,像this这样的glTF小部件定义了可以提供给它的数据类型同样,聊天窗口小部件可能具有自己的输入验证器和输出验证器,但是,诸如聊天窗口之类的那些大而复杂的窗口小部件会在各种应用程序之间重复-电报,whatsapp,线路,whatsapp等- -所有用户都有非常相似的聊天窗口,并创建了一个协议定义的聊天窗口小部件,开发人员会考虑使用聊天应用来显示事物的所有可能需求,以便拖放该小部件的非程序员用户可以拥有很多兼容的数据源,而无需进行任何编程。

[chronological], I agree, we need like a pallete of widgets, or a library, and these widgets must be available globally, so it could be like a browser plugin that brings those widgets into existence, and, I suppose accessible via something like right-clicking of mouse in a new empty browser tab to drop a new widget from a library like

In fact, dead widget snapping is already realized within app prototyping tools for designers, like,, etc.

However, those are not live widgets, and not for end users:

  • too granular (end users generally won't have time to build a menu from scratch, or a chat window from scratch, etc.)
  • not protocol-defining (a complex widget usually defines a composite data type or even interaction protocol itself, for example, a glTF widget like this defines the data type that can be provided to it for it to work properly. Similarly, a chat widget may have its own input validator, and an output validator. However, those big and complex widgets like chat window, repeats across apps -- telegram, whatsapp, line, whatsapp, etc. -- all have very similar chat windows, and creating a protocol-defining chat window widget, the developers would think about all possible needs of chat apps to display things, so that the non-programmer user who drags and drops this widget can have lots of compatible data sources without having to program anything.

insert into widgets (type, name) values ("dropdown menu", "countries") I'd like to be able to define what widgets are available separately from their positioning on the screen. Positioning widgets is harder than deciding what widgets are available. We should separate these functions to make snappy widgets a reality.

    : Mindey
    :  -- 
    :  --