On Sat, Jul 9, 2011 at 7:30 PM, Ashar Voultoiz <hashar+wmf(a)free.fr> wrote: I'm grabbing the BB emulator (BlackBerry® Device Simulators v6.0.0.570 (9780)) now.

Also sprach Ashar Voultoiz: Opera comes in two flavors for mobile devices: Opera Mini and Opera Mobile. Opera Mobile is, indeed, close to the desktop version in the sense that it runs the same display, javascript engine etc. on the device. Opera Mini runs these engines in server parks in the fixed network and tranfers a binary representation to a small viewer on the device. We currently process around 60 billion pages per month and Wikipedia is typically in the top 10 lists in the top 20 countries we publish statistics for:  http://www.opera.com/smw/2011/05/ In the test swarm link you sent, Opera 10 and 11 are listed, but not Opera Mini (which is currently at version 6). Could it be that your sniffer doesn't pick up Opera Mini users?  http://toolserver.org/~krinkle/testswarm/ Here a sample UA string from a recent version of Opera Mini:  Opera/9.80 (Android; Opera Mini/6.24556/25.657; U; en) Presto/2.5.25 Version/10.54 And here's the Opera Mini emulator:  http://www.opera.com/mobile/demo/ While Wikipedia remains popular with Opera Mini users, there is a technical problem which limits the user experience. Wikipedia uses JavaScript to unfold sections in articles. Alas, executing JavaScript requires a rountrip to the server (the Opera Mini server, that is) which takes time and costs money. It would be better if articles were unfolded by default for Opera Mini users.

On Mon, Jul 11, 2011 at 11:40 AM, Brion Vibber <brion(a)pobox.com> wrote: I can tell you right now that Firefox for Android has never redirected to the mobile site for me. If you want more device/version details let me know... Steven _______________________________________________ Wikitech-l mailing list Wikitech-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l

On Mon, Jul 11, 2011 at 12:36 PM, Tomasz Finc <tfinc(a)wikimedia.org> wrote: Mozilla generally recommends using CSS media queries and other client-side techs for adapting your pages to small or large screened-devices; while this is generally a good idea, it doesn't help directly with an issue like this where we'd really prefer to know a binary "device claiming to have a tiny freaky screen" or "anything else" so we can divide people down the mobile-optimized or regular web site paths. (We need to support older/more primitive phones that don't handle any of this stuff.) There are a couple closed-with-extreme-prejudice bugzilla entries like this: https://bugzilla.mozilla.org/show_bug.cgi?id=625238 https://groups.google.com/group/mozilla.dev.platform/browse_thread/thread/4… which mostly look like they're about wanting / not wanting whole gobs of device data in the user-agent string. All *we* really want is "are you a small screen -> include 'Mobile' in the UA" or "otherwise -> don't include 'Mobile' in the UA"... it may or may not be worth seeing if that can get added in as a compatibility thing, however I'm not sure offhand how easy it actually would be to detect whether such a flag should be added or not. I know that iOS has an explicit way to find out whether the app is running in the phone-style UI (iPhone, iPod Touch, and iPhone-targeted apps running on iPad in compat mode) or the tablet-style UI (iPad). I don't know if there's an equivalent on Android. An alternative if that can't be shoehorned in upstream is to do a JavaScript-side check while loading the regular web view; if we're in a browser where CSS media queries detect a tiny mobile screen, and we don't have a redirect preference cookie, then do the redirection after the fact. (And optionally set a default state for the per-browser preference cookie so we only have to do the test once instead of every visit?)

-- brion _______________________________________________ Wikitech-l mailing list Wikitech-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l

----- Original Message ----- Please remember that some people with high-function browsers *want* the low-function results... and some people with *what your code thinks are low-function browsers* want the "standard" results; a user-controllable per-browser persistent cookie to override is a very nice touch here.

TestSwarm 'doesn't need my help' when I load it with Fennec 5 or Opera Mobile 11.1.