I have been trying to determine how the viewport -- and the contents within -- are affected by the viewport meta tag used for drawing content with WebView or with the native browser.
What I have run in to are some apparent inconsistencies. I created a small page (see below) with an image and some javascript for displaying the viewport size so I can visually see the scaling with the image as well as get the exact numbers.
First, a some observations:
Device #1: Physical screen is 1024x600 and it's running Android 2.2.
Device #2: Physical screen is 800x480 and it's running Android 2.3.2.
Of these results, the following don't make sense:
Does anyone know what's going on with those that don't make sense?
Does anyone know why many of the values are 10 pixels shy of match the physical pixels, but that result is as if they match? (e.g. 1.7 and 2.6)
Am I doing something wrong, or does it appear that it's impossible to zoom out beyond 1.0 unless the user is also allowed to scale and the minimum-scale value is set?
That is, even though the valid values are 0.01 to 10, initial-scale values under 1.0 are ignored unless you can also set the minimum scale.
The Android docs do say when user-scalable is no that the min/max scale values are ignored. That doesn't seem very useful. And 2.9-2.11 seem to show that you can't just calculate it yourself and set the width.
Finally, the HTML I'm using:
<html>
<script src="http://code.jquery.com/jquery-1.4.4.js"></script>
<head>
<meta name="viewport" content="initial-scale=0.5, width=device-width, target-densitydpi=device-dpi, minimum-scale=0.1, user-scalable=no" />
</head>
<body style="margin:0;">
<div id="bld" class="bld" style="position:absolute; width:250px; height:125px; left:50px; background-color:#ccccff;">Hello world.</div>
<img id="bl1" src="http://commondatastorage.googleapis.com/kf6nvr/images/1024x768.png" />
<script>
$("#bl1").click(function(){
var pageWidth = $(document).width();
var pageHeight = $(document).height();
var viewportWidth = $(window).width();
var viewportHeight = $(window).height();
$("#bld").html("Page width: "+pageWidth+"<br />pageHeight: "+pageHeight+"<br />port width: "+viewportWidth+"<br />port height: "+viewportHeight);
});
</script>
</body>
</html>
So... what am I missing?
Interesting can of worms you've opened up there. As you try out more devices you'll probably see even more weirdness, and random bugs. Interesting times! :-)
I suspect that somewhere down the line you might want to give up, and just try to...
border-image
s as well as background-image
s coupled with the well-supported background-size
property1 to make things render nice and crisp2.1) To provide backwards compatibility to older browsers (like MSIE8) you'll need to use double background-image
declarations - like:
background-image: url(low-res.png); /* CSS2 */
background-image: url(high-res.png), none; /* CSS3 */
background-size: 100px 10px; /* CSS3 */
2) The -webkit-max-device-pixel-ratio
media query may also help, but as the name implies it only works for webkit browsers.
Ramble:
The new generation Android and iOS devices have such varying screen DPIs that they've resorted to reporting CSS pixels rather than actual device pixels. This is both good or bad - depending on how you look at it.
It's bad because you're not guaranteed anymore the down-to-the-device-pixel control you were used to having, working with the mostly homogenous screens of yore.
On the other hand, it's good because you know your designs will never render so tiny that they can't be read/used by regular humans.
The problem with using target-densitydpi=device-dpi
is that it while it works wonders on a 7inch wide 800px screen, it will completely ruin your application on a 4inch wide 960px screen.