- From: Patrick H. Lauke <redux@splintered.co.uk>
- Date: Tue, 6 Sep 2016 17:34:09 +0100
- To: public-mobile-a11y-tf@w3.org
On 06/09/2016 17:24, Jonathan Avila wrote: >> As 1 inch=25.4mm, so 6px per mm on the tablet, although that is 2D >> which I think means 2.5px per non-square mm. Therefore 20mm wide = >> 50px on that device. > > I think this statement is looking the wrong way. It may be that on > that device 20mm = 50px -- but that doesn't mean 50px is going to be > sufficient on other devices because it won't be 20mm. If they say > 20mm is what was needed then we'd need to figure out how many CSS > pixels would 20mm be on a target device. Are we seriously retreading this conversation? There is no reliable way for a developer to determine this. A developer cannot check the actual physical screen size of a device, or the actual precise metrics like screen DPI, therefore a developer cannot determine with any accuracy what actual physical size something is rendered at on an arbitrary target device's screen. > Just looking at an iPad, a 50px button may be sufficient -- but on an > iPhone 6s the same button is considerably smaller -- likely too small > for many. Do you have an actual test page where you verified this? And does the page set the ideal viewport of width=device-width? > I wonder if we could somehow create a relative size > requirement with some minimum. Like the size of a control in device > independent pixels must be no smaller than 1/15 of the smallest > viewport dimension with a minimum of 50 device independent pixels. How would a developer/auditor test this with any reliability to ensure it's true across devices? The reason for choosing CSS pixels when viewport is set to width=device-width is that it's reliably testable, AND shifts the onus of having a sensible ideal viewport on devices/user agents. Adding a relationship to whatever the device's viewport is actually set means that pass/fail will depend on the exact device the developer/auditor is using? Assume that an auditor is using a device/user agent which, for whatever reason, decided - against what most other mainstream devices/browsers do - to adopt a strangely high viewport dimension ... this would result in a fail on that particular device for a control, while the same control would pass if the auditor used another device. P -- Patrick H. Lauke www.splintered.co.uk | https://github.com/patrickhlauke http://flickr.com/photos/redux/ | http://redux.deviantart.com twitter: @patrick_h_lauke | skype: patrick_h_lauke
Received on Tuesday, 6 September 2016 16:34:28 UTC