Implemented named arguments

[donaldharvey]

I've reimplemented named arguments using the same syntax as James' earlier implementation - i.e. .mixin(@arg: value, @arg2: value).

Named arguments are particularly useful when working with CSS frameworks. For instance, if you had a mixin like this:

.grid(@columns:12) {
    .col(@cols: 1, @offset-left: 0, @offset-right: 0, @vertical-margin: 0, @horizontal-margin: 0, @vertical-padding: 0, @horizontal-padding: 0, @float: left) {
        width: (@cols * 100% / @columns) - (@horizontal-padding * 2) - (@horizontal-margin * 2);
        padding: @vertical-padding @horizontal-padding;
        margin: @vertical-margin @horizontal-margin @vertical-margin @horizontal-margin;
        margin-left: @offset-left * 100% / @columns + @horizontal-margin;
        margin-right: @offset-right * 100% / @columns + @horizontal-margin;
        float: @float;
    }
}

You can then either use it like this, as usual:

.wrapper {
    .grid(12);
}

.sidebar {
    .grid .col(6);
}

Or set specific parameters like this:

.wrapper {
    .grid(12);
}

.sidebar {
    .grid .col(
        @cols: 6,
        @offset-left: 1,
        @horizontal-margin: 2%
    );
}

I'm not yet convinced that this is the best syntax: it's certainly consistent with mixin definitions, but it feels a little bit strange if you're passing variables in with the same name as the keyword argument (e.g. .mixin(@width: @width)). Feedback and/or suggestions on an alternative syntax from LESS users would be appreciated!

The implementation places constraints on keyword arguments similar to those of Python: errors are thrown when positional arguments are placed after keyword arguments, when multiple values are received for a parameter, and when unexpected keyword arguments are received.

Finally, I'm not really sure how well named arguments work with combined with some of the recent additions to mixins (mainly patterns and @arguments); the current behaviour works, but might be unintuitive. Again, feedback from users would be helpful!

[donaldharvey]

@cloudhead, what are your thoughts on this? Is there anything preventing this from being pulled?

[joeldrapper]

+1 this would make mixins so much more powerful! I quite like the syntax you've used as it's the same as you use to specify defaults if you were creating a mixin.

[donaldharvey]

Closing this as James' updated branch, which does this, was just merged in.