A UI toolkit for the Canvas API inspired by UIKit for iOS.
If you know of Zebkit, this is similar but has "nicer" support for mobile UI idioms (see below).
This is a set of UI elements/controls/widgets/views that together comprise a "toolkit" for making user interfaces for games or apps. It relies on just the Canvas API and thus works in places like HTML5 canvas or in native implementations of the Canvas API such as Ejecta.
Most users that want a UI for Canvas use a DOM element placed atop the canvas. I had a situation where I wanted to avoid a browser-based environment (due to performance issues for a game) but had an accelerated canvas API -- namely Ejecta. Thus, without a DOM and all that comes with it, I was left to create a UI toolkit for canvas that worked in or out of a browser environment.
Yes, this is a pretty niche scenario but perhaps this is useful to a few people anyway.
- Follows basic ideas of UIKit from iOS which has been pretty successful
- View system is really easy to understand, use, and extend
- View layout is done in code but is simple and works well at various device resolutions given convenience layout helpers like "place this view in corner", "place this view above that view", etc. No interface builder, insane auto-layout constraint system, hard-coding magic numbers, etc.
- Redraw is on-demand not continuous (doesn't drain device battery if nothing is happening)
- Animations via tweening on any view properties
- Hooks/callbacks/events for most things
- Canvas based apps/games
- Ejecta based iOS apps/games
- Rapid prototyping
- Internationalization and localization are required (e.g. right-to-left text)
- Text selection is needed
- Copy and paste is needed
- Multi-line text editing is needed
- A true "native" experience is paramount for a mobile target
- Hierarchical view system with clipping, shadows, borders, etc.
- Simple code-based view layout
- Touch, mouse, and keyboard handling
- Single, double, arbitary taps; long press
- Built-in drag and drop to/from any view
- Partial transparency
- Tween-based animations
- Label
- Button
- Image view with various content modes
- Modal views (alert, confirm)
- Scrolling with content view, interia scrolling, indicators
- Progress bar
- Slider
- Notification/toast
- Segment/choices
- Navigation stack
- Table view w/ cells, data source, delegate, header/footer
- Left swipe to delete in table views
- Long press and drag to reorder table view cells
- Table view search/filtering
- Word-wrapping for multiline text display with scrolling
- A custom single-line text input view with basic Emacs and OS X shortcuts (extensible)
- A software or on-screen keyboard ala iOS (QWERTY)
- Paging/carousel
The code is written in ES2015 and is to be transpiled for browser usage (or elsewhere like Ejecta) via Babel as most browsers at this time (late 2015) do not support ES2015 directly.
Build your app by require
-ing the Canvas UI sources you need. Then transpile
with Babel, bundle all of it together with Browserify, and distribute the ES5
bundle.
Install Node.js (and with it NPM).
Install dependencies:
$ cd canvas_ui
$ npm install
Symlink the Canvas UI src directory (canvas_ui/canvas_ui
) into node_modules
so that any demo or application code can just require('canvas_ui/xyz')
.
$ cd node_modules
$ ln -s ../src canvas_ui
See this for more details on this technique.
To automatically transpile to ES5, bundle, and reload the updated code, I use:
Here's how:
$ npm install budo garnish -g
$ cd demo
$ budo ./src/demoMain.js:demoBundle.js --live -- -t babelify | garnish
Then open your browser to http://localhost:9966. Note that the demo/index.html
has a script
tag loading demoBundle.js
.
Edit any of the demo files or Canvas UI files and save. They will be babelified via browserify via budo and the browser will be refreshed.
If you hit an exception and make edits while the exception is stuck in the dev tools (e.g. "Pause on Caught Exceptions" is checked in Chrome), you will need to manually refresh the browser post-save.
To build the demo (replace with your own application as needed) for deployment:
$ browserify -e src/demoMain.js -t babelify -o demoBundle.js
$ npm install -g uglify-js
$ uglifyjs demoBundle.js -c -m -o demoBundle.min.js
Then update the script
tag for demoBundle.min.js
.
UglifyJS2 is used a minifier of the bundled ES5 code.
TODO:
- mangle properties too via UglifyJS2. the concern is that the text edit actions use strings; I haven't tried this at all yet.
On redraw, the root view on down is redrawn. Dirty-rect clearing and redrawing only what is changed seemed like overkill given that a lot of work would still need to be done to correctly composite without cached bitmaps of each already-rendered view.
There's no support for multi-touch and gesture recognizers. There is built-in support for single, double, n-tap; long-press; dragging/panning. What's missing from say iOS is: pinch, swipe, rotate. Note that there is support for swiping table view cells that are editable. This could be generalized.
See the TODO
file.
Brian Hammond brian@fictorial.com
MIT