| 07:47 | <smaug____> | hsivonen: ping |
| 09:15 | <jgraham> | New theory: document.write is only the second worst part of the platform and window.open is the worst |
| 09:16 | <Ms2ger> | Can I enter HTML serialization as a contestant? |
| 09:17 | <Ms2ger> | If I have a DOM like <div><input><span></span></input></div>, what is div.innerHTML? |
| 09:19 | <jgraham> | If I window.open a window and there is an alert in the opener window, what happens in the new window? What about if the user has explicitly navigated the opened window? |
| 09:35 | <zcorpan> | Ms2ger: <input> |
| 09:36 | <Ms2ger> | Per spec and in Opera, yes |
| 09:36 | <zcorpan> | do all other browsers disagree? |
| 09:36 | <Ms2ger> | Gecko has "<input><span></span>" |
| 09:36 | <Ms2ger> | Chrome has "<input></input>" |
| 09:37 | <zcorpan> | i don't object to changing the spec what gecko has. chrome's serialization seems weird though |
| 09:37 | <Ms2ger> | I would object to changing to what Gecko has :) |
| 09:37 | <zcorpan> | why? |
| 09:38 | <Ms2ger> | Because I think it's rather weird :) |
| 09:39 | <zcorpan> | not weirder than non-text nodes in <style> |
| 09:39 | <Ms2ger> | Hmm, IE10 seems to match Gecko |
| 09:39 | Ms2ger | looks at what style does |
| 09:40 | <zcorpan> | we can't make innerHTML be "correct" for all cases because the dom can contain lots of things that are not serializable |
| 09:40 | <Ms2ger> | Yeah |
| 09:47 | Ms2ger | wanders off for a bit, before innerHTML gives him a headache |
| 14:50 | <annevk> | todo: add event handler attributes to http://wiki.whatwg.org/wiki/Howto_spec |
| 14:59 | <annevk> | context: http://lists.w3.org/Archives/Public/public-device-apis/2012May/thread.html#msg283 |
| 21:39 | <asmodai> | I long for a day where user agent strings are meaningful again |
| 21:44 | <Philip`> | The only reason for them to be meaningful is if you're trying to extract meaning from them, in which case you're presumably trying to use that meaning to distinguish browsers and handle them differently in some way, in which case a future browser will benefit from lying and saying something that it doesn't really mean but that produces the desired behaviour |
| 21:44 | <Philip`> | so your longed-for day will not be a stable equilibrium |
| 21:45 | <Philip`> | so there will be no long-term gain and plenty of short-term pain |
| 21:45 | <Philip`> | so we are doomed :-( |
| 21:47 | <asmodai> | heh |
| 21:48 | <espadrine> | wow. That was the most perfect answer possible. |
| 21:50 | <Philip`> | (The only stable equilibrium is when every browser is identical in every way that web pages can distinguish, so the user agent string must convey zero bits of information, e.g. by being the union of every UA string used by every other UA in the history of the web) |
| 21:51 | <wilhelm> | Speaking of useless headers – we should do a purge of both request and response headers. |
| 21:53 | <asmodai> | In my case with meaningful I just meant name + version and no using it for anything special. |
| 21:54 | <Philip`> | What's the point of knowing that information if you're not going to use it? |
| 21:55 | <asmodai> | Philip`: Only for seeing adoptation rates and such. |
| 22:46 | <gsnedders> | wilhelm: annevk was trying |