W3C home > Mailing lists > Public > www-dom@w3.org > October to December 2010

Re: [DOM Level 3 Events] MouseEvent.getCoordsAt(element)

From: Jonas Sicking <jonas@sicking.cc>
Date: Tue, 5 Oct 2010 21:29:45 -0700
Message-ID: <AANLkTinaoyv90tYuPovk8eZL76Nz1RzZUsQ1WSnJHgxg@mail.gmail.com>
To: Jonathan Watt <jwatt@jwatt.org>
Cc: www-dom@w3.org
On Tue, Oct 5, 2010 at 3:06 AM, Jonathan Watt <jwatt@jwatt.org> wrote:
> Hi WG,
>
> Background:
>
> Positioning elements based off the position of a pointer event is regularly
> tripping people up in SVG, and with the advent of CSS-transforms, I imagine also
> in HTML.
>
> To position an element based off the position of a pointer event you need to get
> the coordinates of the event in the local coordinate system of the element. To
> do that currently authors have to get the *correct* element-to-root transform,
> invert it, create an SVGPoint, copy the event coordinates to the SVGPoint, then
> use the inverted transform to transform the point and read back the coordinates.
> For something so simple, it's not obvious that this is what you need to do, or
> even how to do it.
>
> Proposal:
>
> To make life easier for authors I'd like to propose the addition of a
> 'getCoordsAt' method to the MouseEvent interface. This method would be passed
> the element at which the local coordinates of the event are required, and return
> an object implementing the SVGPoint interface[1] (or a new interface
> implementing 'x' and 'y' properties).
>
> I've uploaded a JavaScript implemented demo of this method here:
>
>  http://jwatt.org/svg/tmp/mouse-relative-positioning.svg
>
> See also the comments in the source code.
>
> Regards,
> Jonathan
>
> [1] http://www.w3.org/TR/SVG11/coords.html#InterfaceSVGPoint

This would this also take account for things like parent plain HTML
elements being scrolled and things like that right?

What happens if you pass in an element which the mouse isn't actually
over? Would it be possible to transform the coordinates to that
elements coordinate space anyway? Or would it simply return a
null-like value? I would hope for the former.

/ Jonas
Received on Wednesday, 6 October 2010 04:56:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 22 June 2012 06:14:06 GMT