Looks cool!

Quick note: it looks like keyboard navigation with 'tab' can still take you to other text-entry fields etc in the background page and let you interact with them on the keyboard while the modal's open.

Should these also open as a full-screen overlay style on small-screen mobile devices (phones)? If so they may need to be fixed to handle scrolling through anything that's taller than a couple lines of text, as on small phones there's very little room once the keyboard's open.


(And as a general note for Flow UI, "Hide" is very ambiguous; does it hide for just me or for everyone? Who is going to read the text I'm entering?)

-- brion

On Mon, Oct 20, 2014 at 9:16 AM, Shahyar Ghobadpour <sghobadpour@wikimedia.org> wrote:
Hey everyone,

We have developed a modal component for Flow, based on the design spec for MediaWiki-UI modals/dialogs. Thus, we have named it mw.Modal.

A very basic and feature-incomplete version of this modal can be seen at: http://en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow (hover over ellipsis icon in topic, click "hide")

This modal implements the design outlined here, but automates a lot of the visual aspects of it based on the input you give it.

Below is an almost-verbatim copy of the original email I sent to another mailing list, (discussion seen here: https://lists.wikimedia.org/mailman/private/e2/2014-September/002494.html ), detailing the general concepts and use:


First, an example using event binding:
// Instantiate mw.Modal
var modal = new mw.Modal(); 
// Open up a modal with HTML
modal.open( "<button>Bye!</button>" ); 
// Bubble click to wrapper, use it to close the parent modal of the target
$( modal.getContentNode() ).bind( 'click', 'button', function () { mw.Modal.closethis ); } );


Visuals (markup and CSS):
As it stands, you can see the current designs for modals here, with a screenshot of a Flow use case. This is being done with some very simple markup: 
flow-ui-modal
flow-ui-modal-layout
flow-ui-modal-heading (optional)
content + quick navigators (automatic)
flow-ui-modal-content
content

The CSS for this works back to IE6. It automatically centers on the viewport, and scales up to max the viewport width/height, at which point the modal would start showing scrollbar(s). No JS is needed for any of this.

Automatic enhancement:
  • Shows title at top if it exists.
  • Shows X on left of title to close if title exists.
  • Shows ✓ form submission shortcut if title exists, and form with a single primary button exists in content.

Caveat: This does not support off-center placement of modals (eg. placing a modal by a specific element). For those use cases, I imagine mw-ui-tooltip would be the solution, and it was written with that in mind.


Use (JavaScript):
A constructor would be called (modal = new mw.Modal( [ String name, ] [ Object settings ] )), which would give you access to control the flow of your modal.
name may be omitted, but is simply there so you can easily determine which modal is currently active with mw.Modal.getModal.
settings can have the following keys: open (same as modal.open), and title.
  1. modal.open( [ Array|Object|Element|jQuery|String contents ] ) - You can visually render the modal using this method. Accepts Element/jQuery, or HTML String. Returns mw.Modal instance.
    • Multi-step modals with an Array. You can pass [ Element, Element ] to have two steps.
    • Multi-step modals with an Object to have named step keys. Pass { steps: [ 'first', 'second', 'foobar' ], first: Element, second: Element, foobar: Element } for three steps.

  2. modal.close() - Closes and destroys the given instance of mw.Modal.
    • (Global) mw.Modal.close( [ Element node ] ) - Closes the currently-open modal (no args), or the modal in which node belongs. Returns false if none, true if successfully destroyed.

  3. modal.go( int|String to ) - For a multi-step modal, goes to a specific step by number or name. Returns false if invalid step, mw.Modal instance otherwise.
    modal.next() and
    modal.prev() - For a multi-step modal, goes to the previous or next step, if any are left. Returns false if none, mw.Modal instance otherwise.
    • (Global) mw.Modal.go( int|String to, [ Element node ] ) - Go to step on the currently-open modal (no args), or the modal in which node belongs. Returns false if none, mw.Modal instance otherwise.
    • (Global) mw.Modal.next( [ Element node ] ) - Next/prev step on the currently-open modal (no args), or the modal in which node belongs. Returns false if none, mw.Modal instance otherwise.

  4. modal.setTitle( String title ) - Changes the title of the modal.

  5. modal.addSteps( Array|Object|Element|jQuery|String contents ) - Adds one or more steps, using the same arguments as modal.open. May overwrite steps if any exist with the same key in Object mode. Returns mw.Modal instance.

  6. modal.setStep( int|String to, Element|jQuery|String contents ) - Changes a given step. If String to does not exist in the list of steps, throws an exception; int to always succeeds. If the given step is the currently-active one, rerenders the modal contents. Returns mw.Modal instance.
    Theoretically, you could use setStep to keep changing step 1 to create a pseudo-multi-step modal.

  7. modal.getSteps() - Returns an Object with steps, and their contents, eg. { steps: [ 1, 'foo', 'bar', 4 ], 1: Element, 'foo': Element, 'bar': Element, 4: Element }
    • (Global) mw.Modal.getSteps( [ Element node ] ) - Get an Array of steps on the currently-open modal (no args), or the modal in which node belongs. False on failure.

  8. modal.getNode() - Returns the modal's wrapper Element, which contains the header node and content node.
    modal
    .getContentNode()
     - Returns the wrapping Element on which you can bind bubbling events for your content.
    • (Global) mw.Modal.getContentNode( [ Element node ] ) - Returns the wrapping content Element on the currently-open modal (no args), or the modal in which node belongs. False on failure.

  9. modal.setInteractiveHandler( String name, Function callback ) - See "inline event handlers" below. Some helper handlers for modals are predefined, including "modalClose", "modalNavigate" (with data-mwui-modal-to attribute), "modalPrev", and "modalNext".

    An additional attribute, data-mwui-interactive-target is a jQuery selector which allows you to point that event to interact with another element, without having to put this selector in your JS. It supports an unorthodox selector, the "closest parent" selector: "<".
    eg. 
    data-mwui-interactive-target="< .foo .bar" will find the closest parent .foo, and then any nodes within .bar.

  10. (Global) mw.Modal.getModal() - Returns an mw.Modal instance if one is active, false otherwise.

  11. modal.getName() - Returns the modal's name, as defined in initialization.

mw.Modal events:
mw.Modal also inherits ooJS' EventEmitter. So, you may bind to individual instances of mw.Modal, or even the global mw.Modal class itself. See wiki. Current events list: open(mw.Modal instance), close(mw.Modal instance, Element target), navigate(mw.Modal instance, String nextStepKey), render(mw.Modal instance) (only triggered on the first render of a given piece of content, not subsequent ones).  Returning false from any of these event callbacks will cancel the event from continuing.


Inline event callers:
You can use special data- attributes to trigger/listen for events without having to specifically bind to those elements. This allows you to write more dynamic HTML, without having to adjust JavaScript when you change elements or selectors. Some predefined helper handlers already exist (see #8 above).

Example: 
<button data-mwui-interactive-handler="modalClose">Close</button>
 
These event handling techniques are borrowed from the way Flow currently "binds" events via markup. We do this, because we found a significant amount of code was dedicated to finding elements with selectors in JS. Any changes to markup required a change to the JS as well. This method allows us to change the markup without changing the JS in many cases, as the selectors are localized to the HTML. In addition, cumbersome parent selecting is handled with our own "closest" selector (<).

In addition, we've been experimenting with an event forwarding mechanism which "forwards" mouse & keyboard events from within the modal, out to the element which triggered the modal to begin with (or any other arbitrary node). This allows us to capture events within the modal, without having to bind directly to it.


--Shahyar


_______________________________________________
Design mailing list
Design@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/design