| 00:00 | <Hixie> | heh |
| 00:00 | annevk | booked a hotel for tomorrow... |
| 00:00 | annevk | arrives in Seattle at 2PM |
| 00:00 | <Dashiva> | "Basically all the heavy lifting for HTML5 had already been done by the WHAT-WG, which wasn\'t bound by vendors as much." |
| 00:00 | <Hixie> | tomorrow? |
| 00:00 | <Hixie> | oh, right |
| 00:00 | <Hixie> | Dashiva: missed that |
| 00:01 | <annevk> | I'm staying at HomeStead which should be pretty close to Microsoft and was the cheapest I could fine (145 USD...) |
| 00:02 | <annevk> | and now I'm checked in too |
| 00:02 | <Hixie> | i arrive on tuesday around noon |
| 00:03 | <annevk> | k |
| 00:06 | <Hixie> | http is so much more complicated than necessary |
| 00:06 | <tantek> | Hixie - 1.0 or 1.1? |
| 00:06 | <Hixie> | 1.1 |
| 00:06 | <Hixie> | comments in header fields |
| 00:06 | <Hixie> | q-values in the TE: header |
| 00:07 | <Hixie> | i mean sheesh |
| 00:08 | <Dashiva> | Sensible specs would take all the fun out of implementing them |
| 00:09 | <Hixie> | also, the way the spec is organised is a mess |
| 00:09 | <Hixie> | it really should split request and response headers into two sections |
| 00:10 | <annevk> | they're splitting the spec up in seven specs it seems with httpbis |
| 00:10 | <Hixie> | and far too many thigns are SHOULDs that should be MUSTs |
| 00:10 | <Hixie> | really? |
| 00:10 | <Hixie> | seven? |
| 00:10 | <Hixie> | that'll be exciting |
| 00:10 | <annevk> | http://lists.w3.org/Archives/Public/ietf-http-wg/2008AprJun/0595.html |
| 00:10 | <annevk> | for instance |
| 00:12 | <annevk> | it seems rather confusing to me, but this is what it seems they're doing |
| 00:15 | <Hixie> | actually that organisation does make sense |
| 00:16 | <annevk> | oh yeah, i'd imagine it makes sense, as one the authors of HTTP made it up, it's just confusing to have to browse through all those separate documents rather than just one |
| 00:17 | <Hixie> | i need to make little stickers that say "Warning: 110" and stick them on food that i throw away |
| 00:17 | <Hixie> | or maybe stick it on bread that i give back to waiters at restaurants when the bread is stale |
| 00:19 | <Philip`> | Hixie: Do you have many HTTP experts eating your trash? |
| 00:19 | <Hixie> | i live in the bay area, so quite possibly |
| 00:25 | <Hixie> | section 4.4 of http 1.1 says, in part: |
| 00:25 | <Hixie> | "This media type UST NOT be used unless the sender knows that the recipient can arse it" |
| 02:52 | <Hixie> | man i wish the http rfc didn't keep talking about what servers did wrong |
| 02:52 | <Hixie> | it's quite distracting |
| 02:54 | <Hixie> | as far as i can tell, an HTTP server that just immediately closes the connection before reading a single byte from the client is "conditionally conforming" |
| 03:04 | <bradee-oh> | lol |
| 05:52 | <Hixie> | another example of the silly overcomplexness of http: |
| 05:52 | <Hixie> | Comparison of expectation values is case-insensitive for unquoted |
| 05:52 | <Hixie> | tokens (including the 100-continue token), and is case-sensitive for |
| 05:52 | <Hixie> | quoted-string expectation-extensions. |
| 06:10 | <gsnedders> | Hixie: yeah, it's fun :) |
| 06:44 | <philipj> | by default simple events in HTML 5 are cancelable, but does anybody have an idea about what canceli |
| 06:44 | <philipj> | cancelling an event like timeupdate (HTMLMediaElement) might mean? |
| 06:46 | <philipj> | would cancelling a volumechange event mean setting the volume back to the old value? |
| 06:52 | <Hixie> | philipj: canceling an event that has no default action has no effect |
| 07:00 | <philipj> | ok, so that they are cancelable is not practically relevant, still I assume that they should be made cancelable to comply with the spec |
| 07:02 | <hsivonen> | the Web in violation of Web Architecture again, film at 11 |
| 07:02 | <Hixie> | hsivonen: haha |
| 07:03 | <Hixie> | philipj: yeah. at some point the spec may be made more sensible about what's cancelable and what isn't |
| 07:03 | <gsnedders> | How can the web be in breach with its own arch.? Oh, wait, this is what happens if you try and spec an overall arch. |
| 07:04 | <hsivonen> | gsnedders: I don't think it is a necessary consequence of trying to spec it |
| 07:04 | <gsnedders> | hsivonen: It contradicts itself too much I think to be spec'd, then whenever you try and spec anything more you'll just find more contradictions |
| 07:04 | <Hixie> | the web doesn't have a coherent architecture |
| 07:04 | <Hixie> | anyone who's tried working on a web browser can see that |
| 07:05 | <gsnedders> | I haven't, but I can see that :P |
| 07:05 | <Hixie> | it is best to just swallow that and move on, imho :-) |
| 07:05 | <Hixie> | man can you imagine if the html5 spec was full of paragraphs like: |
| 07:05 | <Hixie> | Many older HTTP/1.0 and HTTP/1.1 applications do not understand the |
| 07:05 | <Hixie> | Expect header. |
| 07:06 | <gsnedders> | "Um, so?" |
| 07:06 | <Hixie> | (given that this spec is nearly 10 years old, these paragraphs are really getting annoying) |
| 07:06 | <gsnedders> | There seem to be a few servers that still send HTTP/0.9 responses to any request |
| 07:07 | <gsnedders> | Which is kinda odd |
| 07:07 | <gsnedders> | XHR is bizarre in that case |
| 07:07 | <philipj> | speaking of web legacy, does anybody know where the default 300x150 box for misc elements comes from? was that the size it happened to be in old netscape or something? |
| 07:08 | <hsivonen> | gsnedders: what servers send 0.9 responses? |
| 07:08 | <hsivonen> | gsnedders: are those servers full of security holes, too |
| 07:08 | <gsnedders> | hsivonen: I dunno. Philip` gave a list of pages that did. |
| 07:08 | hsivonen | wonders who profides security updates for the Netscape Enterprise Server |
| 07:09 | <gsnedders> | hsivonen: http://krijnhoetmer.nl/irc-logs/whatwg/20080606#l-818 |
| 07:09 | <Hixie> | philipj: yeah, i think netscape used that for iframes or ilayers or something, and that's where it comes from |
| 07:09 | <hsivonen> | a friend of mine inhereted the admin duties of a site that has been in existence for quite a while, and some of the content had <plaintext> in the file system assuming 0.9 serving |
| 07:09 | <gsnedders> | hsivonen: Oddly enough using cURL I don't get that |
| 07:09 | <hsivonen> | but the server software had long since been updated |
| 07:10 | <gsnedders> | http://www.guadeloupe-fr.com/ doesn't work at all in cURL |
| 07:11 | <hsivonen> | gsnedders: interesting |
| 07:11 | <hsivonen> | gsnedders: I wonder if these server really date from the 0.9 era |
| 07:11 | <gsnedders> | the rest seem to be Apache/1.3 with mod_layout |
| 07:11 | <gsnedders> | And work in cURL, yet not in the HTTP client in Java Philip` used |
| 07:11 | <hsivonen> | gsnedders: or if there's just an incompetently written script copying stuff to stdout that goes straight to a socket |
| 07:12 | hsivonen | hasn't encountered mod_layout |
| 07:12 | gsnedders | opens up telnet |
| 07:12 | <hsivonen> | gsnedders: www.guadeloupe-fr.com doesn't work with telnet for me |
| 07:12 | <gsnedders> | Nor me |
| 07:13 | <gsnedders> | How bizarre. |
| 07:13 | <gsnedders> | Works fine in Safari |
| 07:13 | <hsivonen> | gsnedders: the drmarin.galeon.com server is not a 0.9 server |
| 07:13 | <gsnedders> | yeah, the rest aren't |
| 07:13 | <hsivonen> | gsnedders: I sent a request with a bogus Host header, and I didn't get the same content back |
| 07:14 | <hsivonen> | gsnedders: so the request dispatcher is virtual host-aware |
| 07:14 | <hsivonen> | it's just the output part that sucks |
| 07:14 | <gsnedders> | and that's HTTP/1.1 |
| 07:14 | <gsnedders> | The rest are sucky output |
| 07:14 | <gsnedders> | But it shows that support is still needed (yay) |
| 07:43 | <gsnedders> | Hixie: How are we meant to get the computer to breath for the sarcasm end tag? |
| 07:52 | <hsivonen> | gsnedders: it's not black box testable |
| 08:06 | <othermaciej> | Hixie: HTML URL does have an amusing acronym |
| 08:06 | <othermaciej> | given the details, calling it a HURL might not be so bad |
| 08:26 | <gDashiva> | At one point I tried to backronym ORLI but I never got anything good |
| 09:21 | <Hixie> | so this idea of mcarter's is great but the cookie idea is only going to fly if the same port is used and if we ignore the scheme change |
| 09:21 | <Hixie> | and the authorisation thing isn't going to work unless we massively complicate the handshake to also include a 401 step |
| 09:24 | <othermaciej> | naturally, it has to be same-port (barring a mechanism like access-control) |
| 09:45 | <Hixie> | http://www.w3.org/mid/g4a63l$tkg$1⊙ggo |
| 09:45 | <Hixie> | some people are surprisingly rude on mailing lists |
| 09:45 | <Hixie> | (probably including me) |
| 09:59 | <Windstoss> | I have a question concerning sessionStorage and localStorage… it is said, that a storage event is fired whenever data is changed to all HTMLDocuments that can access the data. How can a sessionStorage have multiple instances that affect each other? Whenever a new window is opened, the session storage area is copied but will no longer affect the original. |
| 10:00 | <Hixie> | Windstoss: generally, it can't |
| 10:00 | <Hixie> | Windstoss: hoever |
| 10:00 | <Hixie> | Windstoss: say you were on page A in a window |
| 10:00 | <Hixie> | Windstoss: and you go to page B at the same domain in the same window, and the browser keeps page A around, just in case you go back |
| 10:00 | <Hixie> | Windstoss: now when you hit back, the page A is shown back on the screen without being reloaded |
| 10:01 | <Hixie> | Windstoss: so now, if you changed anything in the sessionStorage while B was showing, the event will be fired on A when you return to it |
| 10:02 | Hixie | wonders if any browsers implement RFC2817 |
| 10:03 | <Windstoss> | Hixie: I see, thanks! |
| 10:03 | <Hixie> | np |
| 10:03 | <Windstoss> | Hixie: Maybe a note would help to clear this up? |
| 10:04 | <Hixie> | eventually there'll be a whole introduction section for the storage apis |
| 10:04 | <Hixie> | actually i guess there already is. but it will probably be longer. |
| 10:05 | <Windstoss> | One more question: when an HTMLDocument is affected, but not active yet, are those events kind of held back until the HTMLDocument is active? |
| 10:14 | <Hixie> | yeah |
| 10:14 | <Hixie> | it should say that somewhere |
| 10:14 | <Hixie> | can't see where right now |
| 10:14 | <Hixie> | but it's there somewhere, unless i removed it accidentally at some point |
| 10:18 | <Hixie> | so what HTTP method should we pretend to use for web sockets |
| 10:18 | <Hixie> | GET? |
| 10:22 | <hsivonen> | SOCKET? |
| 10:22 | hsivonen | has no idea what he is talking about |
| 10:22 | <Hixie> | heh |
| 10:25 | <hsivonen> | ooh. the WHATWG blog dashboard has a new appearance |
| 10:25 | <hsivonen> | "there are 107 comments in your spam queue right now" |
| 10:27 | <hsivonen> | all real spam |
| 10:27 | <hsivonen> | deleted |
| 10:30 | jgraham | deleted the 7 articles apparently advertising holidays in Spain |
| 10:30 | <hsivonen> | jgraham: I though I just deleted them |
| 10:30 | <hsivonen> | thought |
| 10:31 | <jgraham> | hsivonen: Maybe we both did :) |
| 10:34 | <Lachy> | did you delete the offending users as well? |
| 10:36 | <hsivonen> | Lachy: in the article case, yes. In the comment cases, no. |
| 10:47 | <Hixie> | notes on web sockets: http://www.whatwg.org/specs/web-apps/current-work/notes |
| 10:47 | <Hixie> | mcarter: see ^ |
| 10:48 | <Hixie> | i used GET instead of OPTIONS, to surprise unsuspecting servers less |
| 10:51 | <Hixie> | the cookies and authentication headers are going to be difficult to use in practice unless the same port is used (i'll throw in some stuff to make them be sent even though the scheme differs) |
| 10:52 | <Hixie> | i intend to drop the rfc 2817 ideas, since regular TLS is good enough i think |
| 10:52 | <Hixie> | and nobody seems to do 2817 anyway |
| 10:52 | <Hixie> | i want to have the server echo the origin and the socket uri as part of the handshake |
| 10:53 | <Hixie> | to (a) check for cross-site access being ok and (b) check that the host header is correctly interpreted |
| 10:53 | <Hixie> | (if the server just echoes them without checking them, then too bad) |
| 10:54 | <Hixie> | anyway, if anyone has comments, drop them here, i'll read them in the morning |
| 10:54 | <Hixie> | bed time now |
| 12:12 | <takkaria> | half the problem with the old guard is that they seem to believe that defining error recovery for something is tantamount to defining a new standard |
| 12:14 | <Lachy> | takkaria, in many cases, it is defining a new standard :-) |
| 12:14 | <hsivonen> | undefined error handling is where vendor lock-in is |
| 12:15 | <takkaria> | Lachy: yes, but in that case all implementations also define a new standard |
| 12:16 | <hsivonen> | aside: it seems that Gecko and WebKit run scripts differently when it comes to dynamic DOM insertions |
| 12:16 | <hsivonen> | the code I announced on the blog has the right nested document.write semantics in WebKit, but apparently not in Gecko |
| 12:19 | <gDashiva> | takkaria: Also, defining error recovery is encouraging errors |
| 12:19 | <takkaria> | gDashiva: yes, that's the one |
| 12:19 | <takkaria> | because if you don't define it, then no-one's interoperable, so no-one will come to rely on any errorr-handling behaviour because it varies so wildly |
| 12:26 | <hsivonen> | takkaria: cunning plan :-) |
| 12:27 | <takkaria> | I suspect it might actually work somewhat if the marketplace was more varied |
| 12:27 | <Lachy> | takkaria, that sounds surprisingly like the mentality of some other working groups I know |
| 12:28 | <takkaria> | e.g. if, say, no one vendor had more than 10% market share |
| 12:29 | <Lachy> | (I'm pretty sure there's a few mails on www-html or www-html-editor from an XHTML2WG members saying roughly the same thing.) |
| 12:34 | <takkaria> | I guess the mentality is a left-over from physical engineering standards |
| 12:35 | <takkaria> | there's not really any such thing as specified error recovery from getting the wrong track width for railways or different-sized screwheads |
| 12:39 | <Lachy> | takkaria, but there does need to be error recovery when some idiot uses the wrong size screw for the job, even if the screw itself was really good quality. |
| 12:40 | <takkaria> | yeah, but the error recovery is specified by the laws of material science and those laws are fairly unchangeable. :) |
| 12:41 | <takkaria> | as least AIUI |
| 13:17 | <roc> | error handling would still be a problem even if no vendor had more than 10% market share |
| 13:17 | <roc> | because you can be sure that *some* error situations would be handled the same way by the majority of engines, just by luck |
| 13:18 | <roc> | so they would become the de facto standard |
| 13:18 | <roc> | and so it goes |
| 16:11 | <hdh> | hsivonen: I host a V.nu at home; the --stylesheet argument's first letter is cut off, it becomes '<link href="ttp://hdh.dyn-o-saur.com/style/v.nu.css"; rel="stylesheet">' |
| 16:19 | <hdh> | line 596 should have been stylesheet = arg[13:] |
| 20:14 | <hsivonen> | hdh: fixed thanks. (the numbers for stylesheet and microsyntaxes were reversed) |
| 20:46 | <Hixie> | mcarter: yt? |
| 21:03 | <hsivonen> | http://tech.slashdot.org/tech/08/06/30/1845201.shtml |
| 21:04 | hsivonen | wonders if commenters are up to speed about canvas and svg |
| 21:13 | <othermaciej> | 2D vector graphics, we'll get right on that, sir |
| 21:16 | <othermaciej> | looks like openajax.org is down, too bad, I was curious what the list is |
| 21:16 | <Hixie> | the list isn't too bad, but it aligns pretty closely with what we've already got in the works, and things we know we should have in the works but don't have editors for |
| 21:16 | <Dashiva> | How about anti-slashdotting support in browsers? :) |
| 21:16 | <Hixie> | some good ideas, though |
| 21:16 | <Philip`> | http://google.com/search?q=cache:http://www.openajax.org/runtime/wiki/Feature_Requests_Summary_Page&strip=1 |
| 21:16 | <Hixie> | i spend more time crafting the checkin comments than making the edits when i'm doing editorial work |
| 21:16 | <Hixie> | that's just silly |
| 21:18 | <Philip`> | http://azarask.in/blog/post/contextfreejs-algorithm-ink-making-art-with-javascript/ wants canvas getTransform() |
| 21:19 | <othermaciej> | a lot of the things they list are things that are already in browsers afaik |
| 21:19 | <othermaciej> | well, some of them anyway |
| 21:20 | Philip` | notices that DNS is already being widely abused to form words, disregarding the whole hierarchy idea, so it seems sensible to allow arbitrary words without forcing a weird dot in them |
| 21:51 | <csarven> | Hixie Would you be able to verify the accuracy of this: The purpose of the <object> element is to allow the browser to run an external application for a non-native data type (e.g., Java applet). |
| 21:51 | <hsivonen> | precise terminology and defined error handling isn't exactly what RSS 2.0 is known for |
| 21:52 | <csarven> | or hsivonen :) |
| 21:59 | <mcarter> | Hixie, whats up |
| 22:50 | <Hixie> | mcarter: http://krijnhoetmer.nl/irc-logs/whatwg/20080630#l-326 |
| 22:56 | <Loughes> | Any Opera developers in here? I have to say that I love the new browser. I am finally looking forward to where the WWW may be going. |
| 23:13 | <mcarter> | Hixie, oh cool, you've gotten started on the WebSocket stuff |
| 23:13 | <Hixie> | indeed |
| 23:14 | <annevk> | weather is nice outside in Seattle |
| 23:14 | <Dashiva> | The talk on www-html about @style is... interesting |
| 23:14 | <annevk> | well, Redmond / Bellevue |
| 23:18 | <Philip`> | Dashiva: It's fortunate that @style got added back into HTML5 :-) |
| 23:19 | <Dashiva> | Philip`: What makes it even more odd is that they recently decided to let @target back into their new XHTML 1.1 SE thingie |
| 23:20 | <Hixie> | mcarter: any feedback on the quoted stuff? (sorry my site appears to be down, I'm working on getting it back up) |
| 23:21 | <jacobolus> | Hixie: whatwg.org seems to be down |
| 23:21 | <jacobolus> | Hixie: so it's hard to see what http://www.whatwg.org/specs/web-apps/current-work/notes actually says :) |
| 23:23 | <mcarter> | Hixie, i think I partially loaded that document, but then the server started hanging... after you get it working I'll give you some feedback |
| 23:23 | <Philip`> | http://pastebin.ca/1059505 is what http://www.whatwg.org/specs/web-apps/current-work/notes said this morning GMT |
| 23:24 | <Philip`> | (No idea if it's changed since then) |
| 23:25 | <Hixie> | that's the one |
| 23:25 | <Hixie> | no change from that |
| 23:25 | <Hixie> | jacobolus, mcarter: ^ :-) |
| 23:25 | <jacobolus> | yep. :) |
| 23:25 | Hixie | has narrowed it down to NFS hanging |
| 23:27 | <mcarter> | Hixie, this looks pretty good at a first glance -- let me get back to you this evening with mroe substantial feedback |
| 23:28 | <Hixie> | k |
| 23:28 | <mcarter> | Hixie, I'm not sure if it makes sense to require the headers to be in an exact order |
| 23:29 | <jacobolus> | Hixie, mcarter: this looks pretty good to me. exciting! |
| 23:32 | <Hixie> | mcarter: the idea is to have a handshake that cannot be faked by tricking the server into sending back certain fields |
| 23:32 | <annevk> | hmm, are we going to introduce new URI schemes too? |
| 23:33 | <annevk> | fun |
| 23:33 | <Hixie> | annevk: it's a new protocol, so i guess yes |
| 23:33 | <Philip`> | URL schemes! |
| 23:36 | <annevk> | seems whatwg.org is up again |
| 23:36 | <Hixie> | indeed |
| 23:46 | <Hixie> | hsivonen: yt? |
| 23:47 | <Hixie> | hsivonen: in http://www.whatwg.org/specs/web-apps/current-work/#open there is an "act as if", but it's unclear what i should link to for that one. I mean, the token would cause all manner of things to happen (e.g. it implies <html>, <head>, <body>, quirks mode, and then finally gets treated as a <pre>) |
| 23:55 | <Hixie> | i wonder what i should hyperlink in these "act as if" cases |