Skip to content

Instantly share code, notes, and snippets.

@jonathantneal
Created November 30, 2011 20:59
Show Gist options
  • Save jonathantneal/1410787 to your computer and use it in GitHub Desktop.
Save jonathantneal/1410787 to your computer and use it in GitHub Desktop.
innerWidth / innerHeight tests @ http://sandbox.thewikies.com/orientation/
innerWidth / innerHeight tests @ http://sandbox.thewikies.com/orientation/
--------------------------------------------------------------------------------
Tested (14 devices, 28 browsers):
Droid 2 Global Android 2.2
iPhone 4 iOS5 (Safari, Opera Mini)
Motorola Atrix Android 2.3.4 (Stock browser, Dolphin, Skyfire, Opera Mini, Firefox)
Samsung Galaxy S9000 Android 2.3 (Webkit, Opera Mobile)
Samsung Galaxy Y Android 2.3.5
Nokia N9 (MeeGo) (Webkit, Opera Mobile)
Palm Pre Plus webOS 2.1.0 Build 284
Palm Pre 2 (WebOS 2.1) (Webkit, Opera Mobile)
Galaxy 10.1 Android 3.2 (Webkit, Dolphin, Opera Mini)
HP TouchPad
iPad 1 iOS5 (Safari, Opera Mini)
iPad 2 iOS5
Motorola Xoom Android 3.2.1 (Stock browser, Dolphin, Skyfire, Firefox)
================================================================================
"p" describes the result of loading the test with the device portrait oriented (taller).
"l" describes the result of loading the test with the device landscape oriented (wider).
"p,l" describes the result of loading the test with the device portrait oriented and then rotated to landscape.
"l,p" describes the result of loading the test with the device landscape oriented and then rotated to portrait.
================================================================================
Droid 2 Global Android 2.2 (handheld):
Web browser (webkit):
meta-none: p 800x1220 - l 800x335
meta-width: p 320x488 - l 569x239
meta-width-initial: p 320x488 - l 569x239
meta-width-initial-maximum: p 320x488 - l 569x239
--------------------------------------------------------------------------------
iPhone 4 iOS5 (handheld):
Safari:
meta-none: p 980x1091 - l 981x425
meta-width: p 320x356 - l 320x139
meta-width-initial: p 320x356 - l 480x208 - p,l 320x139
meta-width-initial-maximum: p 320x356 - l 480x208
Opera Mini:
meta-none: p 320x416 - l 480x256
meta-width: p 320x416 - l 480x256
meta-width-initial: p 320x416 - l 480x256
meta-width-initial-maximum: p 320x416 - l 480x256
* note: opera mini does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
Motorola Atrix Android 2.3.4 (handheld):
Stock browser:
meta-none: p 960x418 - l 800x1241
meta-width: p 360x559 - l 640x279
meta-width-initial: p 360x559 - l 640x279
meta-width-initial-maximum: p 360x559 - l 640x279
Dolphin:
meta-none: p 800x1161 - l 960x364
meta-width: p 360x523 - l 640x243
meta-width-initial: p 360x523 - l 640x243
meta-width-initial-maximum: p 360x523 - l 640x243
Skyfire:
meta-none: p 800x1176 - l 800x351
meta-width: p 360x529 - l 640x281
meta-width-initial: p 360x529 - l 640x281
meta-width-initial-maximum: p 360x529 - l 640x281
Opera Mini:
meta-none: p 318x450 - l 565x203
meta-width: p 318x450 - l 565x203
meta-width-initial: p 318x450 - l 565x203
meta-width-initial-maximum: p 318x450 - l 565x203
* note: opera mini does not reflow the layout on orientation change, so these numbers are not changed.
Firefox:
meta-none: p 800x1366 - l 800x418
meta-width: p 360x615 - l 640x335
meta-width-initial: p 360x615 - l 640x335
meta-width-initial-maximum: p 360x615 - l 640x335
* note: firefox does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
Nokia N9 (MeeGo) (handheld):
Web (webkit):
meta-none: p 980x1519 - l 978x465
meta-width: p 320x496 - l 567x269 - p,l 320x152 - l,p 140x270
meta-width-initial: p 320x469 - l 567x269 - p,l 320x152 - l,p 140x270
meta-width-initial-maximum: p 320x496 - l 567x269 - p,l 320x152 - l,p 140x270
Opera Mobile (old preview):
meta-none: l 488x184
meta-width: l 488x184
meta-width-initial: l 488x184
meta-width-initial-maximum: l 488x184
* note: Opera Mobile (old preview) does not have portrait mode
--------------------------------------------------------------------------------
Palm Pre Plus webOS 2.1.0 Build 284
web v2.1.0 (webkit):
meta none: p 960x1400 - l 960x1400
meta width: p 320x452 - l 320x1400
meta width initial: p 320x452 - l 320x320
meta width initial maximum: p 320x452 - l 320x320
* note: web does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
Palm Pre 2 (WebOS 2.1) (handheld):
Stock browser:
meta-none: p 960x1400 - l 960x1400
meta-width: p 320x452 - l 320x452
meta-width-initial: p 320x452 - l 320x452
meta-width-initial-maximum: p 320x452 - l 320x452
--------------------------------------------------------------------------------
Samsung Galaxy S9000 Android 2.3 (handheld):
Web browser (Webkit):
meta-none: p 800x1130 - l 800x357
meta-width: p 320x452 - l 533x239
meta-width-initial: p 320x452 - l 533x239 - p,l 320x452
meta-width-initial-maximum: p 320x452 - l 533x239
Opera Mobile:
meta-none: p 981x1232 - l 980x347
meta-width: p 320x403 - l 534x190
meta-width-initial: p 320x403 - l 533x189
meta-width-initial-maximum: p 320x403 - l 533x189
* note: opera mobile does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
Samsung Galaxy SCP (GT-I9003) Android 2.3.4 (Gingerbread) (handheld):
Stock Browser
meta-none: p 800x1130 - l 800x358 - p,l 800x1130 - l,p 800x1130
meta-width: p 320x452 - l 533x239 - p,l 320x452 - l,p 320x452
meta-width-initial: p 320x452 - l 533x239 - p,l 320x452 - l,p 533x239
meta-width-initial-maximum: p 320x452 - l 533x239 - p,l 533x239 - l,p 533x239
--------------------------------------------------------------------------------
Samsung Galaxy Y Android 2.3.5 (handheld):
Web browser (webkit):
meta-none: p 800x853 - l 800x440
meta-width: p 320x341 - l 427x235
meta-width-initial: p 320x341 - l 427x235
meta-width-initial-maximum: p 320x341 - l 427x235
--------------------------------------------------------------------------------
Galaxy 10.1 Android 3.2 (tablet):
Web browser (Webkit):
meta-none: p 980x496 - l 980x1392
meta-width: p 800x1128 - l 1280x648
meta-width-initial: p 800x1128 - l 1280x648
meta-width-initial-maximum: p 800x1128 - l 1280x648
Dolphin:
meta-none: p 980x1389 - l 980x506
meta-width: p 800x1141 - l 1280x661
meta-width-initial: p 800x1141 - l 1280x661
meta-width-initial-maximum: p 800x1141 - l 1280x661
Opera Mini:
meta-none: p 471x698 - l 753x415
meta-width: p 471x698 - l 753x415
meta-width-initial: p 471x698 - l 753x415
meta-width-initial-maximum: p 471x698 - l 753x415
* note: opera mini does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
HP TouchPad - webOS 3.0.4 Build 77 (tablet):
web v3.0.4 (webkit):
meta none: p 3072x3768 - l 4096x2744
meta width: p 768x3768 - l 1024x2744
meta width initial: p 768x642 - l 1024x686
meta width initial maximum: p 768x942 - l 1024x686
* note: web does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
iPad 1 iOS5 (tablet):
Safari:
meta-none: p 980x1185 - l 981x644
meta-width: p 768x928 - l 768x504
meta-width-initial: p 768x928 - l 1024x672 - p,l 769x505
meta-width-initial-maximum: p 768x928 - l 1024x672
Opera Mini:
meta-none: p 768x956 - l 1024x700
meta-width: p 768x956 - l 1024x700
meta-width-initial: p 768x956 - l 1024x700
meta-width-initial-maximum: p 768x956 - l 1024x700
* note: opera mini does not reflow the layout on orientation change, so these numbers are not changed.
--------------------------------------------------------------------------------
iPad 2 iOS5 (tablet):
Safari:
meta-none: p 980x1185 - l 981x644
meta-width: p 768x928 - l 768x504
meta-width-initial: p 768x928 - l 1024x672 - p,l 769x505
meta-width-initial-maximum: p 768x928 - l 1024x672
* note: these results are identical to the iPad 1.
--------------------------------------------------------------------------------
Motorola Xoom Android 3.2.1 (tablet):
Stock browser (webkit):
meta-none: p 980x1382 - l 980x496
meta-width: p 800x1128 - l 1280x648
meta-width-initial: p 800x1128 - l 1280x648
meta-width-initial-maximum: p 800x1128 - l 1280x648
Dolphin:
meta-none: p 980x1392 - l 980x502
meta-width: p 800x1136 - l 1280x656
meta-width-initial: p 800x1136 - l 1280x656
meta-width-initial-maximum: p 800x1136 - l 1280x656
Skyfire:
meta-none: p 800x1146 - l 980x534
meta-width: p 800x1146 - l 1280x697
meta-width-initial: p 800x1146 - l 1280x697
meta-width-initial-maximum: p 800x1146 - l 1280x697
* note: this browser has an odd quirk. If you start in landscape & switch to portrait, it refreshes the values. If you start in portrait and change to landscape, it won't change. You must refresh.
Firefox:
meta-none: p 980x1509 - l 980x576
meta-width: p 800x1232 - l 1280x752
meta-width-initial: p 800x1232 - l 1280x752
meta-width-initial-maximum: p 800x1232 - l 1280x752
* note: firefox does not reflow the layout on orientation change, so these numbers are not changed.
================================================================================
results collected by @stefsull, @drublic, @pascalvos, @alex_gibson, @tchalvak, and @jon_neal
@stefsullrew
Copy link

Yea, I just showed that to @jonathantneal a couple hours ago. Too funny! (Just have been too tired to compare the data from both. Maybe tomorrow.)

@cubiq
Copy link

cubiq commented Dec 2, 2011

@stefsullrew
Copy link

Heh. Yea, Jonathan & I talked about that earlier... I think with all the data we've collected (with more to come), it's more digestible in spreadsheet form. Wat do you think @JonathanNeal? (Well, you can answer when you wake up. ;))

@alexgibson
Copy link

I discovered some weird behaviour on the BlackBerry Playbook (1.0.8.4985)

Just testing the meta viewport alone has some odd results (note meta-width test in landscape is wacko):

meta none: p 1024x1598 - l 1024 x 512
meta width: p 600 x 936 - l 600 x 300
meta width initial: p 600 x 936 - l 1024 x 512
meta width initial maximum: p 600 x 936 - l 1024 x 512

I noticed this was not the same as other site's I've built & tested on the Playbook. I discovered that if you include the BB proprietary HandheldFriendly rule <meta name="HandheldFriendly" content="True" />, then the meta viewport test results are consistent.

Include: <meta name="HandheldFriendly" content="True" /> plus meta viewport

meta none: p 1024x1598 - l 1024 x 512 (HandheldFriendly content="False")
meta width: p 600 x 936 - l 1024 x 512
meta width initial: p 600 x 936 - l 1024 x 512
meta width initial maximum: p 600 x 936 - l 1024 x 512

Just another day building for mobile devices, sigh!

@quimcalpe
Copy link

Samsung Galaxy SCP (GT-I9003)
Android 2.3.4 (Gingerbread)
Stock Browser

  • meta-none: p 800x1130 - l 800x358 - p,l 800x1130 - l,p 800x1130
  • meta-width-initial-maximum: p 320x452 - l 533x239 - p,l 533x239 - l,p 533x239
  • meta-width-initial: p 320x452 - l 533x239 - p,l 320x452 - l,p 533x239
  • meta-width: p 320x452 - l 533x239 - p,l 320x452 - l,p 320x452

@cubiq
Copy link

cubiq commented Dec 2, 2011

I was thinking to a way to automate the data collection.
I've set up a quick demo page which automatically collects all screen dimensions by asking the user to rotate the device.
The result is a json file like this http://cubiq.org/dropbox/width/collect.php that can be a displayed a little better like so http://cubiq.org/dropbox/width/results.php

The data collection itself is at http://cubiq.org/dropbox/width/ (mobile device only!). What do you think? Better ideas on how to export the JSON?

@tchalvak
Copy link

tchalvak commented Dec 2, 2011 via email

@cubiq
Copy link

cubiq commented Dec 2, 2011 via email

@tchalvak
Copy link

tchalvak commented Dec 2, 2011 via email

@cubiq
Copy link

cubiq commented Dec 2, 2011 via email

@jonathantneal
Copy link
Author

I'll be back at lunch to update this based on all of your contributions. Thanks everyone!

@jonathantneal
Copy link
Author

@kurtcoda and @quimcalpe - your results were added.

@jamesgpearce, yes, i hope this data correlates with your results. I was unaware of other projects, and I had no idea there would be this kind of contribution from the community.

@cubiq and @stefsullrew, yes, I agree that a spreadsheet of this data would be much more useful. I also agree that automating the data collection would be ideal, but like @tchalvak mentioned, I would want to make sure the automation doesn't allow for faulty results due to which orientation the device was loading the page with.

@cubiq, I'd be up for refining the project as long as we all knew and shared our goals for what to do with the results. This project has far exceeded the scope I was originally looking for, but that's no reason to deny the wealth of information this could provide for so many others. One thing that bothers me is how unreadable these charts could become. I favor storing the data json somehow, because with json we can create all sorts of readable and visually stimulating charts.

@jamesgpearce
Copy link

It's very cool. Can I also suggest we consider beaconing the data into BrowserScope?

@cubiq
Copy link

cubiq commented Dec 2, 2011

@jonathantneal, I can share the source code and release it under MIT license it. Please consider that it's a 15 minutes projects, the output at http://cubiq.org/dropbox/width/results.php is just a quick demo, data should be presented into a table not a list. Also we should double check if data is collected properly and also I'm not completely convinced by the JSON structure.

If the device is orientation aware the process is bullet proof. Otherwise we should make a more detailed step by step procedure asking the user to rotate the device in the correct position. False positives can be avoided collecting more results for the same device.

Data is currently stored into a sqlite db, the JSON file you see at http://cubiq.org/dropbox/width/collect.php is a duplicates-free output.

Data collected should be freely accessible to anyone, would be nice to set up an API where you can query the DB by user-agent and get back the right screen size.

@stefsullrew
Copy link

I've wanted to crowd source this info for the past year—ever since Greg and I started testing our devices and finding ridiculous differences. In the past year, another zillion new devices have come out — each with their own quirks. I just A) don't own all devices ;) and B) don't have the skillz to really make it automated (and Greg's too busy).

That said, I would LOVE it if we could build something whereby people with devices could run the tests, the info is stored and then retrievable by many. I think it's important for accuracy to have more than one test on a device (who knows what a variety of different setup/OS/etc will do). I know from my own testing that reports vary by browser on the same device, so it's important to record the browser as well as the OS and device.

I'm "all in" in any way I can assist in this project as it unfolds. :) We own an iPad 1, iPad 2, (2) Galaxy Tab, (2) Galaxy 10.1, Xoom, Playbook with one of the newer Amazon/Nook type tablet we'll be getting shortly. Handhelds are Droid2, Nexus One, Blackberry Torch, iPhone 3G, Atrix, Galaxy S II (I think that's what it is, it's with Greg now)... Maybe something else, can't remember. All that's to say, that's all I can personally know about without something like this. But if we crowd source & compile, we at least have data to make decisions by (though there's nothing like really viewing your project on a real device).

@stefsullrew
Copy link

@cubiq - I just ran my Atrix through your test... Two things I see if this is developed. We're going to have to have a way to put in what type of mobile it is, yes? The Atrix reports things oddly (even found a thread about it here: http://www.tera-wurfl.com/forum/viewtopic.php?f=2&t=96) ... also, you can change your user agent string in Android pretty easily. It would be a shame to have someone change it, not note that somewhere, and report something incorrect.

Same thing with the actual browser. Since most of the Android browsers are webkit based, we'll need to have a way to specify default browser vs Dolphin vs Skyfire (or any others). I found a list someone has going here: http://www.zytrax.com/tech/web/mobile_ids.html But good lord it's all over the place (and for instance, for Dolphin it says the strings are scattered about the manufacturer's website... So maybe grab the string AND let them report as well?

OK, that's my brain dump for now...

@kurtcoda
Copy link

kurtcoda commented Dec 3, 2011

@cubiq - I just tried to ran your app on my webOS devices but had no luck so far. So these user-agent strings might be useful in case the devices are of any interest.

palm Pre Plus:

Mozilla/5.0 (webOS/2.1.0; U; en-GB) AppleWebKit/532.2 (KHTML, like Gecko) Version/1.0 Safari/532.2 Pre/1.1

HP TouchPad:

Mozilla/5.0 (hp-tablet; Linux; hpwOS/3.0.4; U; en-GB) AppleWebKit/534.6 (KHTML, like Gecko) wOSBrowser/234.76 Safari/534.6 TouchPad/1.0

@cubiq
Copy link

cubiq commented Dec 5, 2011

An easy solution for all browsers is to ask the user to start in portrait > rotate > tap to confirm > rotate > tap to confirm > and so on. Maybe we could also check if the device is orientation aware and in that case send the confirmation tap automatically.

Regarding user agent spoofing there's little we can do about it. We could ask what browser/device the user is on at the beginning (trying to auto suggest our best bet looking at the user agent). This would mean making a list of all possibile combinations. Or you want the user to type his/hers browser and device?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment