v2 checklist

[Adam Pearce]

• insert
• ƒ.call and ƒ.not - still not sold on their implementation; why set d3.f.not = function(d){ return !d } instead of calling with a string?
• translateX/translateY? Kind of annoying to add selection and transition for each
• change loadData to not take an array
• add links to source in docs
• reorder docs - group selection modifiers, v3 functions and everything else?
• docs for parent
• okay to remove bower.json?
• okay to use an adult license?
• Manually add d3 dependencies to package.json

Missing functions

selection.prop

jetpack added selection.prop as alias for selection.property. Much faster to type, isn't it? Also only consistent with selection.attr, and familiar to jQuery folks.

// for everyone's sake, let's add prop as alias for property
d3.selection.prototype.prop = d3.selection.prototype.property;

selection.on

jetpack lets you set the same listener for multiple events at once, jQuery style.

selection.on('click touchend', function() {
    console.log('this works on desktop AND mobile!');
});

// this tweak allows setting a listener for multiple events, jquery style
var d3_selection_on = d3.selection.prototype.on;
d3.selection.prototype.on = function(type, listener, capture) {
    if (typeof type == 'string' && type.indexOf(' ') > -1) {
        type = type.split(' ');
        for (var i = 0; i<type.length; i++) {
            d3_selection_on.apply(this, [type[i], listener, capture]);
        }
    } else {
        d3_selection_on.apply(this, [type, listener, capture]);
    }
    return this;
};

[Gregor Aisch]

• not sure who added bower.json, probably not me
• yes, adult license is ok
• translatex/translatey -- alternatives? something like .translate(d3.f([['value', x], 0]))
• what does "instead of calling it with a string mean again? you mean instead of doing d3.f('value', 'f.not')?

[Gregor Aisch]

actually, I did create bower.json. yes, fine to remove. jetpack is baked into d3 builds nowadays

[Adam Pearce]

The array notation for ƒ looks interesting... but I'm not sure how we'd indicate that in .translate(d3.f([['value', x], 0])) the 0 is a constant, not d => d[0]. I think pretty soon we will be using fat arrows, so maybe .translate(d3.f('value', x, d => d[x, 0]))will suffice?

Otherwise, we could stick a bunch of utility functions on d3.f:

• d3.f.not = function(d){ return !d }
• d3.f.asXpos = function(d){ return [d, 0] }
• d3.f.asYpos= function(d){ return [0, d] }
• d3.f.run = function(d){ return d() }
• d3.f.clamp = function(min, max){ return function(d){ return Math.max(min, Math.min(d, max)) } }

Sticking random functions on ƒ seems like less API overhead than attaching them to the selection.prototype - you'll know that they're just simple functions and not have to guess what they do.

The magic string in d3.f('value', 'f.not') makes me a little uncomfortable... if there is ƒ.not function you could use it for other things: numZeroScores = scores.filter(ƒ.not)

[Gregor Aisch]

ok, we can kill magic string functions. and I'm fine with the utility functions, but I use them rarely, so we can also throw them away and I just add them if I need to.