| 00:13 | <cying> | is there an event fired when a CSS WebFont has finished loading? |
| 00:18 | <hober> | cying: you might look at how google's font loader does it |
| 00:19 | <cying> | hober: *nod* they do it by polling a hidden div for a size change every 50 ms |
| 00:19 | <cying> | hober: but i was hoping there was a more cool way of doing that |
| 00:24 | <billh> | how call mercurial for dev html5lib (python) |
| 00:25 | <karlcow> | Philip`: 49 tests in 10 min. |
| 00:25 | <Rik`> | cying: there's not at the moment but they are some discussions to add one |
| 00:26 | <cying> | Rik`: that would be awesome and very helpful |
| 00:29 | <Philip`> | karlcow: Sounds fun |
| 00:29 | <Philip`> | In the olden days when Safari couldn't automatically check the canvas output and I had to run the tests manually, I did something like 1 per second but still got bored with it |
| 00:29 | <billh> | anyone here know html5lib? |
| 00:30 | <Philip`> | billh: Yes |
| 00:30 | <billh> | I assume the tarkit is rather old |
| 00:31 | <karlcow> | a bit slow, the waiting time in between each tests \o/ Let's hope many people participate |
| 00:31 | <billh> | so the wiki say the dev version can be retrieved with mercurial -- where do I go |
| 00:35 | <Philip`> | billh: http://code.google.com/p/html5lib/source/checkout |
| 00:37 | <Philip`> | Hixie: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10647 probably shouldn't be NEEDSINFO |
| 00:38 | <Hixie> | fixed |
| 00:39 | <Hixie> | but i wonder why it was NEEDSINFO |
| 00:39 | <Hixie> | have i marked lots of other things NEEDSINFO recently? |
| 00:39 | <Hixie> | i wonder if my scripts are failing on the new bugzilla |
| 00:41 | <Hixie> | looks like it goes to NEEDSINFO whenever I go back after bugzilla tells me it didn't know an e-mail address |
| 00:48 | <billh> | Philip: thanks |
| 04:14 | <hober> | ok, I pushed the elisp html5 tokenizer to github if anyone wants to play with it |
| 04:15 | <hober> | I'm in the middle of hooking it up with html5lib's tokenier tests, so I don't actually know if it works beyond ad-hoc testing, but it seems to work. :) |
| 08:58 | <jgraham> | I wonder how hard it would break the web to make <aol> into a parser macro that expands to "Me too" |
| 09:03 | <Hixie> | feedback on bugs 10419 (muted), 9843 (parser), 9895 (img.complete), 10539 (parser), 10743 (intro), 10694 (console), 10672 (video), 9590 (vendor prefix), 10625 (parsing), 10798 (selection), 9429 (<s>), and 10320 (websrt voice) would be most welcome from anyone who sees this, as i'm somewhat blocked on all of those |
| 09:03 | <Hixie> | and without your input i'll just make stuff up :-) |
| 09:03 | <hsivonen> | Hixie: scary |
| 09:05 | <hsivonen> | the styling on http://www.w3.org/Bugs/Public/ is weird |
| 09:06 | <othermaciej> | Hixie: my survey of Apple folks seems to lean in favor of x-vendor-attribute, with corresponding IDL (if any) of elt.vendorAttribute, but so far I only have answers from people who read mail late at night |
| 09:07 | <hsivonen> | I'm compiling my browser in place, so I can't post comments right now, but re: 9843 as of two weeks ago, none of the major browsers blocks document.written scripts on sheets |
| 09:07 | <Hixie> | othermaciej: k |
| 09:08 | <Hixie> | hsivonen: how do they define "document.written script"? one that is run by a nested parser invocation? |
| 09:09 | <Hixie> | i have to go to bed, but if anyone has any feedback on those bugs please put it in the bugs when you have the chance, that way i don't lose it if my irc client dies overnight |
| 09:09 | <Hixie> | thanks in advance |
| 09:10 | <Hixie> | nn |
| 09:22 | <hsivonen> | Hixie: Gecko (as of now) and IE *don't* define document.written scripts based on the nesting level at the time of tokenization. |
| 09:22 | <hsivonen> | oh. Hixie went away already |
| 09:30 | <jgraham> | Sigh. I see that I have entirely failed to be aware of another ISSUE before it reached poll |
| 09:32 | <hsivonen> | jgraham: cleanuptable? |
| 09:33 | <jgraham> | Yeah |
| 09:41 | <annevk> | guess I now need to actually test rightmargin et al |
| 09:43 | <annevk> | I'm pretty sure non-IE browsers only support topmargin/leftmargin |
| 09:43 | <annevk> | has anyone started some kind of Rendering Chapter Test Suite project? |
| 09:47 | <jgraham> | annevk: No |
| 09:47 | <jgraham> | Kind of hard to do because it is all non-normative |
| 09:48 | <annevk> | for typical browsers it's pretty much a must |
| 09:49 | <Philip`> | I thought someone had produced a big table of all the default stylesheet rules of all the browsers, or something like that |
| 09:49 | <jgraham> | Maybe, but "typical browsers [...] pretty much" hides a world of hurt |
| 09:49 | <Philip`> | which could be compared against what the spec expects |
| 09:49 | <annevk> | jgraham, I have no idea what you're talking about |
| 09:50 | <annevk> | jgraham, most of the rules there are there because of compatibility concerns, not because we thought they were funny |
| 09:50 | <annevk> | jgraham, so they need to be tested |
| 09:51 | <jgraham> | I very well understand that. Nevertheless it is very tricky to write tests for non-normative content |
| 09:52 | <annevk> | I don't understand |
| 09:52 | <jgraham> | I don't want people feeling compelled to change their rendering on e.g. small screen devices to get a better score in some test but worse user experience |
| 09:54 | <annevk> | I think you're envisioning problems that are not really there |
| 09:55 | <jgraham> | I am not against these tests existing |
| 09:55 | <jgraham> | But they need big warnings about which UAs they should apply to |
| 09:55 | <annevk> | Anyway, if anything that has nothing to do with the trickyness of writing the tests; just with how they are published/promoted |
| 09:56 | <jgraham> | Well sure, the difficulty is not in testing the guidelines, but in making the difference between these tests and those for actual normative content clear |
| 10:32 | <annevk> | whoa, Minefield does support bottommargin and rightmargin |
| 10:32 | <annevk> | but only in quirks mode |
| 10:36 | <annevk> | but analysis in bug report but did not reopen: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10564 |
| 10:36 | <annevk> | s/but/put/ |
| 13:23 | <jgraham> | http://www.reddit.com/r/reddit.com/comments/dk3s0/the_ie9_team_responds_to_your_questions/ seems to be much less interesting than it sounds |
| 13:24 | <hsivonen> | Is there a canonical summary that explains why the Selector API does with namespace bindings what it does? |
| 13:27 | <annevk> | it doesn't do namespaces |
| 13:55 | <karlcow> | i18n storm |
| 13:56 | <jgraham> | really |
| 13:56 | <Workshiva> | s5m |
| 13:56 | <Peter`> | let's hope they don't have hundreds of comments |
| 13:57 | <Peter`> | though most comments seem ok |
| 13:57 | <karlcow> | that's often the issue with deadlines sometimes. |
| 13:58 | <karlcow> | I expect the Last Call to be also quite interesting. |
| 13:59 | <jgraham> | Some subset of the i18n bugs seem to be UI issues that are strictly out of scope |
| 13:59 | karlcow | realizes that "often" and "sometimes" make a quite awkward sentence. |
| 14:00 | <Philip`> | Where did i18n comment 4 go? |
| 14:00 | <Philip`> | (Also 13 and 20 seem to be in Bugzilla but not on the list) |
| 14:01 | <jgraham> | I have 20 |
| 14:01 | <jgraham> | Not 13 or 4 |
| 14:01 | <jgraham> | Maybe in i18n circles the number 4 is like the number 13 in many western cultures |
| 14:01 | <Philip`> | Okay, I guess 20 is stuck somewhere in my pipes |
| 14:02 | <Workshiva> | Tubes |
| 14:02 | <karlcow> | Philip`: chinese curse :) 4 = number of death |
| 14:02 | <jgraham> | Hotels run by i18n people never have a floor 4 |
| 14:02 | <Workshiva> | So that's why HTML was bumped to 4.01 so fast... |
| 14:03 | <annevk> | way to submit comments at the last possible moment :/ |
| 14:03 | <annevk> | oh well |
| 14:03 | <Philip`> | When is the last possible moment? |
| 14:03 | <karlcow> | http://en.wikipedia.org/wiki/Numbers_in_Chinese_culture#Four |
| 14:03 | Philip` | wonders if he should submit some |
| 14:04 | <karlcow> | Philip`: the moment be4 death |
| 14:04 | <annevk> | Philip`, the pre Last Call date was October 1 |
| 14:04 | <annevk> | Philip`, everything after that will automatically be handled as a Last Call comment |
| 14:04 | <Philip`> | Oh, that's loads of time |
| 14:04 | Philip` | wonders what timezone October 1 is in |
| 14:05 | <karlcow> | if it's Bugzilla timezone, then it is EST |
| 14:05 | <jgraham> | I think October 1st occurs in all Earth timezones |
| 14:06 | <Philip`> | Ah, if it's EST then that's two whole days (rounded to the nearest whole days) so no rush |
| 14:06 | <annevk> | zcorpan, we are removing Cookie2 support fwiw |
| 14:09 | <zcorpan> | annevk: oh |
| 14:27 | <karlcow> | ah 4 is in |
| 14:28 | <annevk> | lots of very specific features |
| 15:04 | <mpilgrim> | annevk: you around? |
| 15:05 | <mpilgrim> | what would be a reasonable JavaScript feature detection script for XMLHttpRequest Level 2? |
| 15:06 | <annevk> | you would have to test it per feature really |
| 15:06 | <mpilgrim> | i'm interested in two features |
| 15:06 | <mpilgrim> | .upload |
| 15:06 | <annevk> | for cross-origin requests what people use is ("withCredentials" in (new XMLHttpRequest)) or some such |
| 15:07 | <mpilgrim> | and the ability to send FormData in the send() method |
| 15:07 | <annevk> | since the whole reason FormData was added was for send() I suppose checking window.FormData is enough |
| 15:07 | <mpilgrim> | ah, excellent |
| 15:08 | <annevk> | upload you can do similarly to withCredentials |
| 15:08 | <mpilgrim> | if ("upload" in (new XMLHttpRequest)) { ... } |
| 15:08 | <annevk> | yeah |
| 15:08 | mpilgrim | wonders if "new XMLHttpRequest" works in all browsers |
| 15:09 | <annevk> | since IE7, I think |
| 15:09 | <annevk> | oh, maybe you want (new XMLHttpRequest()) |
| 15:10 | <paul_irish> | yup |
| 15:10 | <annevk> | I think per ECMAScript it's just a stylistic difference, but I've really only tested the latter widely |
| 15:10 | <mpilgrim> | both variants appear to work in chrome |
| 15:11 | <mpilgrim> | but i'll take your word for it |
| 15:11 | <jgraham> | They should be identical |
| 15:11 | <annevk> | both also work in Gecko and Opera |
| 15:11 | <jgraham> | Provided browsers follow the spec |
| 15:11 | <annevk> | dunno about IE |
| 15:11 | <jgraham> | So, erm, good luck with that ;) |
| 15:13 | <annevk> | mpilgrim, writing a chapter on XMLHttpRequest? |
| 15:13 | <mpilgrim> | no, just adding some feature tests to the appendix |
| 15:13 | <mpilgrim> | someone emailed and requested |
| 15:14 | <paul_irish> | annevk: you know how gecko has a global DataTransfer object? is that specced out? it not being available in webkit is really a downer sometimes. |
| 15:14 | <paul_irish> | mpilgrim: to peeks, pokes? |
| 15:14 | <mpilgrim> | apparently, when you name a list "a guide to detecting everything," some people take it literally |
| 15:14 | <annevk> | mpilgrim, can I request a chapter? :) |
| 15:14 | <annevk> | paul_irish, isn't that part of drag & drop? |
| 15:15 | <paul_irish> | annevk: yup. but hey you're smart.. you might know offhand. :) |
| 15:15 | <annevk> | http://www.whatwg.org/specs/web-apps/current-work/complete.html#the-dragevent-and-datatransfer-interfaces |
| 15:15 | <annevk> | paul_irish, jaja, those tricks only work a few times ;) |
| 15:16 | <annevk> | (I think it's also used for copy and paste, an API which is not defined yet :/) |
| 15:17 | <annevk> | mpilgrim, could be part of the Python book too, how to create a web service that works cross-origin; but maybe it's a little too experimental still |
| 15:17 | <hsivonen> | annevk: wasn't CORS almost done in 2006? |
| 15:18 | <annevk> | hsivonen, changes have been requested several times |
| 15:19 | <annevk> | hsivonen, I think it's now more or less done, but still being tweaked, and not implemented everywhere yet |
| 15:22 | <annevk> | http://security.sys-con.com/node/1544072 -- Crockford flaming again in the media without any kind of constructive suggestion |
| 15:22 | <annevk> | what a tool |
| 15:24 | <micheil> | actually, raising the point is okay even so |
| 15:24 | <micheil> | other wise those posting bug reports in the open are also tools |
| 15:24 | <annevk> | this is not a bug report |
| 15:24 | <micheil> | true |
| 15:24 | <micheil> | but it's still raising a ponit |
| 15:25 | <micheil> | a lot of what I've heard from douglas is about fixing what's broken before adding more that could be broken |
| 15:25 | <annevk> | and if he actually paid closer attention (and the media would ask for different perspectives before going ahead and publishing) he might learn about sandboxing of untrusted content |
| 15:26 | <annevk> | so far he hasn't even done such a thing as filing a bug report, he's just yelling from the sidelines |
| 15:26 | <micheil> | annevk: I guess in the end someone has to do that, right? |
| 15:26 | <mpilgrim> | ok, http://diveintohtml5.org/everything.html updated with 3 "XMLHttpRequest Level 2" tests |
| 15:26 | <annevk> | micheil, for what benefit? |
| 15:27 | <annevk> | micheil, spread some FUD? |
| 15:27 | <micheil> | maybe not much, but there was always that saying: some publicity is better then no publicity. |
| 15:27 | <mpilgrim> | everything who helped answer my questions in the past 5 minutes gets a free copy of the online edition |
| 15:28 | <annevk> | i feel so privileged |
| 15:30 | <mpilgrim> | and the knowledge that you've helped make the print edition even more out of date |
| 15:30 | <annevk> | right, and I paid for that one! |
| 15:31 | <mpilgrim> | the withered ghost of the traditional publishing industry thanks you |
| 15:32 | <jgraham> | annevk: You're clearly too old for standards. You're changing stuff faster than your library can cope with |
| 15:34 | <mpilgrim> | then again, that withered ghost has sold over 12,000 copies of my book |
| 15:35 | <mpilgrim> | like actual sales, with money changing hands and everything |
| 15:35 | <mpilgrim> | people baffle me |
| 15:36 | <annevk> | time for Ghost Busters? |
| 16:10 | <jacobolus> | annevk: is there ever a time where "new foo" does different than "new foo()"? |
| 16:10 | <Workshiva> | Not if the spec has a word in it |
| 16:10 | <annevk> | dunno |
| 16:11 | <jacobolus> | annevk: for instance, I use "+new Date" all the time to get the time as # of milliseconds since the epoch |
| 16:11 | <Workshiva> | Well, obviously it matters if you don't end with a semicolon |
| 16:11 | <Workshiva> | new foo()('/index.html') is very different from new foo('/index.html') |
| 16:12 | <jacobolus> | Workshiva: sure. sometimes (+new Date) |
| 16:12 | <micheil> | Workshiva: but doing stuff like (fab)(router)(/^azs$/) |
| 16:12 | <micheil> | is far cooler. :P |
| 16:12 | <Workshiva> | Also, evil |
| 16:13 | <micheil> | it looks so bizzar |
| 16:13 | <jacobolus> | mpilgrim: so I think "new XMLHttpRequest" *should* be the same everywhere :) |
| 16:13 | <jacobolus> | that is, the same as "new XMLHttpRequest()" |
| 16:15 | <mpilgrim> | but both fail gracelessly in IE6, right? |
| 16:16 | <jacobolus> | something like that |
| 16:16 | <mpilgrim> | excellent |
| 16:17 | <jacobolus> | MSXML2.XMLHTTP, etc. |
| 16:17 | <mpilgrim> | in unrelated news, i finally got an email from someone complaining that i'd removed the sentence "This site is optimized for Lynx just because fuck you." from the front page of diveintopython3.org |
| 16:17 | <Workshiva> | I would like to file a formal objection against the removal |
| 16:18 | <Ms2ger> | Please write a change proposal |
| 16:18 | <Workshiva> | "Call the [[Construct]] method on Result(2), providing no arguments (that is, an empty list of arguments)." |
| 16:18 | <mpilgrim> | i had previous gotten numerous complaints about its presence |
| 16:19 | <mpilgrim> | including MORE THAN ONE from Lynx users who complained that it was not actually optimized for Lynx |
| 16:20 | <Workshiva> | Did you tell them it was optimized for another version than the one they were using? |
| 16:21 | <mpilgrim> | i've found it is unwise to troll Lynx users |
| 16:24 | <Workshiva> | Do they gather their posse and diss your CSS files? |
| 16:24 | <jacobolus> | mpilgrim: you could serve a separate version of the page to lynx user agents with the sentence left out |
| 16:27 | <Philip`> | It's slightly distracting that the body font in diveintohtml5.org makes me read Modernizr as "Modemizr" |
| 16:27 | <Philip`> | which would be a good name for a browser extension that simulates a really slow network connection |
| 16:28 | <jacobolus> | mpilgrim: btw, unless you like them for extra readability, you can leave off parens in many places on http://diveintohtml5.org/everything.html |
| 16:28 | <jacobolus> | mpilgrim: in binds tighter than && for example |
| 16:28 | <jacobolus> | well, in a few places |
| 16:29 | <mpilgrim> | lies! custom fonts make your design look dean and modem! |
| 16:30 | <micheil> | duh. |
| 16:30 | <Workshiva> | Them's fighting words |
| 16:31 | <AryehGregor> | annevk, I wasn't really planning on working out what features of <marquee> should be supported, no, beyond the IDL stuff that I'm working on anyway. |
| 16:31 | <jacobolus> | (and new binds tighter than in, of course) |
| 16:32 | <mpilgrim> | since Lynx doesn't execute script, I could just add the sentence with document.write |
| 16:32 | <tabatkins> | Hmm. An extension that simulates a really slow network connection would be pretty useful, actually. |
| 16:33 | <nimbupani> | tabatkins: http://stackoverflow.com/questions/473465/firefox-plugin-to-simulate-slow-internet-connection-or-limit-bandwidth |
| 16:33 | <tabatkins> | Oh Firefox, is there anything you don't have an extension for? |
| 16:33 | <nimbupani> | mute volume on specific tabs :( |
| 16:33 | <mpilgrim> | jacobolus: i'm willing to streamline the code |
| 16:34 | <mpilgrim> | some concrete examples? |
| 16:34 | <jacobolus> | mpilgrim: return ('localStorage' in window) && window['localStorage'] !== null; |
| 16:34 | <jacobolus> | return ('sessionStorage' in window) && window['sessionStorage'] !== null; |
| 16:35 | <jacobolus> | and then the two at the bottom with (new XMLHttpRequest()); |
| 16:36 | <AryehGregor> | jacobolus, the general rule is 1) multiplication and division bind tighter than addition and subtraction, 2) arithmetic and bitwise operators bind tighter than logical operators, 3) logical operators bind tighter than comparison operators, 4) use parentheses for everything else because otherwise you're just going to confuse people reading the code who haven't spent hours memorizing the language's precedence rules. |
| 16:36 | AryehGregor | hates it when people do things like omit parentheses when they mix && and || |
| 16:36 | <jacobolus> | AryehGregor: fair enough |
| 16:37 | <jacobolus> | AryehGregor: people should learn that && and || bind looser than almost everything in most languages |
| 16:37 | <AryehGregor> | I think "in" is like "+" or something, though, I think it's pretty obvious it binds tighter than "&&". 'localStorage' in (window && ...) doesn't make any sense. |
| 16:37 | <AryehGregor> | Yeah, that's my (2). |
| 16:37 | <jacobolus> | AryehGregor: in javascript the only things looser than && and || are assignment and the ternary conditional |
| 16:38 | <AryehGregor> | Yeah, ternary is looser than basically anything. |
| 16:38 | <jacobolus> | actually I looked at https://developer.mozilla.org/en/JavaScript/Reference/Operators/Operator_Precedence |
| 16:38 | <AryehGregor> | So that you can do a == b ? foo : bar. |
| 16:38 | <jacobolus> | apparently also "comma" is loosest |
| 16:38 | <AryehGregor> | Well, that hardly counts. |
| 16:38 | <AryehGregor> | You're not going to do f(foo, bar + baz) and expect it to mean f((foo, bar) + baz), seriously. |
| 16:39 | <AryehGregor> | Is there any case where there are multiple ways to associate commas and other operators and more than one isn't a syntax error? |
| 16:39 | <jacobolus> | but anyway, if you see foo in bar && baz === spam, it shouldn't be that hard to learn that such things will be (foo in bar) && (baz === spam) |
| 16:39 | <AryehGregor> | Right, that I agree with. |
| 16:40 | <jacobolus> | mpilgrim: !!('something that presumably returns yes or no').replace(/no/, '')); is a cute construction. who came up with that? |
| 16:41 | <mpilgrim> | i think i did |
| 16:41 | <mpilgrim> | and it's not "yes"/"no" |
| 16:41 | <Philip`> | What's wrong with "!=== 'no'"? |
| 16:42 | <mpilgrim> | it's "probably"/"maybe"/"no"/"" |
| 16:42 | <mpilgrim> | and only older versions of firefox ever return "no" |
| 16:42 | <AryehGregor> | jgraham, MS's response in that reddit link links to a log of this channel, did you notice? |
| 16:42 | <AryehGregor> | Although they apparently didn't figure out how to link to specific links. |
| 16:42 | <Philip`> | Link to links? |
| 16:42 | <jgraham> | AryehGregor: No, I think I drowned in the waffle before that |
| 16:42 | <AryehGregor> | They probably meant to link to this: http://krijnhoetmer.nl/irc-logs/whatwg/20100625#l-151 |
| 16:42 | <AryehGregor> | http://www.reddit.com/r/reddit.com/comments/dk3s0/the_ie9_team_responds_to_your_questions/ |
| 16:42 | <AryehGregor> | (for the "why don't we get 100 on Acid3") |
| 16:43 | <mpilgrim> | ok, http://hg.diveintohtml5.org/hgweb.cgi/rev/0c2514cfbd99 |
| 16:43 | <AryehGregor> | "As you can imagine a hospital with a multimillion dollar patient tracking web based application doesn't want a silent or automatic upgrade to their browser that could in fact jeopardize their patient's safety." |
| 16:44 | <AryehGregor> | Why do people always use these emotive and totally improbable examples? |
| 16:44 | <nimbupani> | coz feelings are stronger than rational thoughts. |
| 16:44 | <Philip`> | AryehGregor: If it's merely improbable and not impossible, then it seems like an important concern |
| 16:44 | <tabatkins> | As you can imagine a hospital with a multimillion dollar patient tracking web based application doesn't want a connection to the outisde internet on their computers that could in fact jeopardize their patient's safety. |
| 16:46 | <Philip`> | I imagine hospitals are also interested in having many of their computers actually be usable, and internet connections are important for that |
| 16:46 | <tabatkins> | Computers with connections to outside don't have to be the same as computers with private patient's data in the multimillion doloor patient tracking application. |
| 16:46 | <AryehGregor> | Philip`, according to statistical mechanics, it is improbable but not impossible that all the air in the room you're standing in right now will randomly move to the other side of the room and stay away from your mouth and nose for five minutes, thus causing you to suffocate. Does this particularly concern you? |
| 16:46 | <jacobolus> | eek. usable for what? not like the people inputting patient records need to use the same machines for chatting on IRC |
| 16:47 | <Workshiva> | Applications that depend on a certain IE version could specify that in the compat header |
| 16:47 | <Workshiva> | The header they themselves added, hmm! |
| 16:47 | <jacobolus> | AryehGregor: you're talking about many many many orders of magnitude difference in probability |
| 16:47 | <AryehGregor> | jacobolus, probably at least 10^100. |
| 16:47 | <AryehGregor> | Orders of magnitude, I mean. |
| 16:47 | <AryehGregor> | Not a factor of 10^100, you understand. |
| 16:48 | <Philip`> | tabatkins: If that application has no connection to the outside world, it's not much good if doctors or ambulances etc need to access it remotely |
| 16:48 | <AryehGregor> | A factor of 10^10^100, or actually probably rather more, but that's the ballpark. |
| 16:48 | <tabatkins> | Philip`: That's what VPNs into the hosting server are for. |
| 16:48 | <AryehGregor> | Anyway, I suspect that people who work outside the hospital/financial/etc. industries have an exaggerated view of how important reliability is. |
| 16:48 | <jacobolus> | AryehGregor: the molecules in my roommate could reassemble themselves into a pink unicorn too, for example |
| 16:49 | <AryehGregor> | jacobolus, that would be even less likely. Possibly add another 10^ at the bottom. |
| 16:49 | <Philip`> | tabatkins: In some cases the risk introduced by internet connectivity would be outweighed by the benefits (in convenience, cost, maintainability, etc) of having it, but only because people like Microsoft are careful to keep the risk relatively low by not throwing giant unwanted software updates at everyone |
| 16:50 | <Philip`> | AryehGregor: That doesn't concern me because I could get up and walk to the other side of the room |
| 16:51 | <AryehGregor> | Philip`, no, because the empty space could follow you. |
| 16:51 | <AryehGregor> | Wherever you went. |
| 16:51 | <AryehGregor> | Including if you tried putting on an oxygen mask. |
| 16:51 | <jacobolus> | .... |
| 16:51 | <tabatkins> | The probability really is roughly equivalent. |
| 16:52 | <AryehGregor> | I know someone who works as a programmer in the financial industry. I think I've said this before. He writes bond-trading software that people use to trade billions of dollars. They write everything in C++, and have no automated testing, and he thinks XML's fast failure is obnoxious. |
| 16:52 | <AryehGregor> | They mostly rely on employing really good programmers, I think. |
| 16:53 | <AryehGregor> | (whereas hospitals mostly rely on the fact that if they mess up and a patient dies due to their incompetence, they have liability insurance) |
| 16:53 | <AryehGregor> | As far as the financial industry goes, I mean, there's lots of money at stake, but probably they make the big clients' lawyers sign some limitation of liability agreement. |
| 16:54 | <AryehGregor> | It's not like the NYSE has to worry that if their computers mess up, they'll be liable for a trillion dollars of lost trades or something. |
| 16:54 | <AryehGregor> | Hixie, what's the status of removing the SMIL and/or SVG Fonts requirements from Acid3? Did anything ever come of that? |
| 16:55 | <mpilgrim> | jesus, i leave this channel for 10 minutes and it goes all to hell |
| 16:57 | <AryehGregor> | jgraham, I think the reddit article is pretty interesting, given that it's a usual Microsoft shill piece. |
| 16:57 | <mpilgrim> | hmm, whwat's the status of getting the world to recognize that Acid3 should not be treated as a standards compliance test on a scale of 0 to 100? |
| 16:57 | <AryehGregor> | They do actually answer some hard questions, although they spin the answers very heavily. |
| 16:57 | <AryehGregor> | (e.g., see answer to iamnotobama) |
| 16:58 | <AryehGregor> | mpilgrim, the status is if we remove SMIL and SVG Fonts as a requirement, everyone will score 100/100. Then we just have to make sure no one makes an Acid4. |
| 16:58 | <AryehGregor> | Using physical force, if necessary. |
| 16:58 | <mpilgrim> | damn, i probably shouldn't have registered diveintoacid4.org |
| 16:59 | <AryehGregor> | Man, Unicode smilies are useless. They're just too small to be legible in normal text. |
| 16:59 | <AryehGregor> | ☺ |
| 16:59 | <Philip`> | mpilgrim: "Inline SVG" on http://diveintohtml5.org/peeks-pokes-and-pointers.html has an unmatched ) |
| 16:59 | <AryehGregor> | At least in my fonts. |
| 16:59 | <mpilgrim> | eek |
| 16:59 | <AryehGregor> | $ whois diveintoacid4.org |
| 16:59 | <AryehGregor> | NOT FOUND |
| 16:59 | <Philip`> | Diving into acid sounds painful |
| 16:59 | <jgraham> | mpilgrim: Maybe "dive into acid" should be the punishment for attempting to create acid 4 |
| 16:59 | <jacobolus> | is there an Acid 4? |
| 17:00 | <jacobolus> | can we start suggesting features for it sometime? |
| 17:00 | <Ms2ger> | Not yet |
| 17:00 | <jacobolus> | can we promote reasonable color management to the top of the list? |
| 17:00 | <AryehGregor> | "Heres a question, what do you have to offer over Chrome or Firefox except for a bigger market of popup ads and browser specific viruses?" "Thanks for the excellent double rhetorical question." |
| 17:01 | <mpilgrim> | Philip`: fixed |
| 17:01 | <Philip`> | mpilgrim: Also, 'shadowColor . . . CSS bg, default = "transparent black"' - there is no such CSS colour as "transparent black" |
| 17:02 | <mpilgrim> | it doesn't say it's a CSS color, it says it's a CSS background |
| 17:02 | <Philip`> | As far as I'm aware there is no such CSS background as "transparent black", and anyway it's a colour not a background |
| 17:03 | <Philip`> | (The CSS keyword "transparent" is sort of defined as rgba(0,0,0,0) so it's black already) |
| 17:04 | <Ms2ger> | FWIW, http://www.hixie.ch/tests/evil/acid/004/ |
| 17:06 | AryehGregor | throws rocks at Hixie |
| 17:06 | <AryehGregor> | This kind of glitzy "score X/100" test with a pretty picture really makes people focus on the wrong things. |
| 17:06 | <AryehGregor> | We should develop real test suites and get people to focus on those. |
| 17:06 | <AryehGregor> | You know, ones that actually test details instead of encouraging implementers to implement exactly the bare minimum necessary features to pass the test. |
| 17:07 | <jgraham> | AryehGregor: Happily, we are :) |
| 17:07 | jgraham | will be disappointed if acid 4 is XML-based |
| 17:08 | <AryehGregor> | Okay, so let's not also do Acid4, because then people are just going to focus on Acid4. |
| 17:08 | <Philip`> | The problem is that implementers don't really bother to pass 100% of real test suites |
| 17:08 | <AryehGregor> | Yes, so we should advertise those more widely. |
| 17:08 | <jgraham> | Mixed namespace HTML5 seems much more fun |
| 17:08 | <Philip`> | AryehGregor: Advertising works best when you have something fun to advertise |
| 17:08 | <AryehGregor> | We should get a unified test suite that you can easily run in your browser that prints out the number of tests failed. |
| 17:08 | <Philip`> | Pretty pictures and "score X/100" are fun since random people can run the tests themselves easily |
| 17:08 | <AryehGregor> | (not percentage, since that would start at 80% or more for a lot of features) |
| 17:09 | <jgraham> | Maybe we should make the official testsuite forma pretty picture |
| 17:09 | <AryehGregor> | Yes, we should do that for the real tests. |
| 17:09 | <mpilgrim> | Philip`: you're right, i was misreading the spec. fixed. |
| 17:09 | <jgraham> | But based on tens of thousands of tests |
| 17:09 | <jgraham> | Not a hundred |
| 17:09 | <jgraham> | It will be slow to run though |
| 17:10 | <Philip`> | I suppose there ought to be an interesting way to visualise thousands of boolean values |
| 17:10 | <Philip`> | so that it converges on some nice coherent image as the values tend towards true |
| 17:10 | <AryehGregor> | Hundreds of thousands of tests. |
| 17:10 | <jgraham> | Or becomes functional somehow |
| 17:11 | <Philip`> | (More interesting than just a plain bitmap) |
| 17:11 | <JBark> | Each fail is a dead pixel? |
| 17:11 | <AryehGregor> | Depending on what you call a "test". |
| 17:11 | <Philip`> | (but at least that'd be a place to start) |
| 17:11 | <AryehGregor> | I've written over 15,000 tests, sort of. |
| 17:11 | <AryehGregor> | Although of course it's really a script that makes 15,000+ assertions. |
| 17:11 | <Ms2ger> | BTW, jgraham, are there any approved html tests that use the test harness, apart from those for getElementsByClassName? |
| 17:11 | <Philip`> | for (var i = 0; i < 15001; ++i) assert(i == i); |
| 17:11 | <jgraham> | Yes, I expect us to get to 100,000 |
| 17:11 | <Philip`> | Now I've written more than you |
| 17:12 | <AryehGregor> | Yeah, that's why counting tests doesn't make much sense. |
| 17:12 | <AryehGregor> | Proper test suites are meant to find bugs, not compare browsers. |
| 17:13 | <AryehGregor> | At least in my tests, in more than one case, I have one bug causing 100 failures or more. |
| 17:13 | <Philip`> | Test suites aren't just meant to find bugs, they're meant to get bugs fixed |
| 17:13 | <AryehGregor> | Well, okay. |
| 17:13 | <Philip`> | and comparing browers motivate developers to fix their bugs |
| 17:14 | <Philip`> | *browsers |
| 17:14 | <Philip`> | *motivates |
| 17:14 | <jgraham> | Ms2ger: I don't know |
| 17:14 | <AryehGregor> | Yes, but that's not how the tests are written. If you want to compare browsers fairly, you need to make sure all the tests are independent and comparable in scope. |
| 17:14 | <annevk> | getElementsByClassName tests might end up getting pulled |
| 17:14 | <AryehGregor> | But that's incompatible with thorough testing. |
| 17:14 | <annevk> | if we make it a DOM Core feature |
| 17:15 | <Philip`> | AryehGregor: You don't need to compare browsers *fairly* |
| 17:15 | <AryehGregor> | Well, if you don't, then you create perverse incentives. |
| 17:16 | <jgraham> | Ms2ger: Why? |
| 17:16 | <AryehGregor> | Like "let's fix this bug that doesn't affect authors at all because it causes 2,000 test failures". |
| 17:16 | <AryehGregor> | Of course, test suites always create perverse incentives. |
| 17:16 | <AryehGregor> | Like I found that Gecko and WebKit will refuse to set tabIndex to anything higher than 32k or something. Violates the spec, but is anyone going to run into it, really? |
| 17:17 | <AryehGregor> | But I'll file bugs on it anyway, because it does violate the spec . . . |
| 17:17 | <Ms2ger> | I was wondering if there was anything else I could run |
| 17:17 | <Philip`> | AryehGregor: The goal of the test suite is usually to have browsers pass 100% of it, so if we added a test it's because we want that incentive to exist |
| 17:17 | <jgraham> | Ms2ger: The XHR tests use the test harness |
| 17:17 | <AryehGregor> | I guess. |
| 17:17 | <annevk> | http://tc.labs.opera.com/apis/XMLHttpRequest/ and http://tc.labs.opera.com/apis/EventSource/ |
| 17:17 | <AryehGregor> | We just have to make sure that browsers don't start cheating. |
| 17:18 | <Ms2ger> | Just for HTML, though |
| 17:18 | <AryehGregor> | There will be pressure for that to happen if there's too much pressure to pass all the tests. |
| 17:18 | <jgraham> | Ms2ger: No. I have some unreleased bit |
| 17:18 | <jgraham> | s |
| 17:18 | <AryehGregor> | Because then the goal will change from "fix bugs highlighted by tests" to "pass tests". |
| 17:18 | <AryehGregor> | The same thing happens with benchmarks. |
| 17:18 | <Philip`> | AryehGregor: I guess we could do that by watching how they fix the bugs, and if they do it in a cheating way then update the test suite to break them again |
| 17:19 | <AryehGregor> | How can you watch how they fix the bugs? |
| 17:19 | <Philip`> | Look in their bug trackers |
| 17:19 | <annevk> | AryehGregor, make sure you test all aspects |
| 17:19 | <AryehGregor> | IE9 apparently blatantly cheats on some of the SunSpider tests, in that its performance drops dramatically if you add "true;" in the middle of a loop or "return;" at the end of a function or whatever. |
| 17:19 | <AryehGregor> | Philip`, works for Gecko and WebKit. Others, not so much. |
| 17:19 | <annevk> | AryehGregor, which admittedly is somewhat hard |
| 17:19 | <AryehGregor> | If you write good enough tests, though, it's true that the easiest way to pass will hopefully be to just fix the bugs. |
| 17:20 | <AryehGregor> | Unless they cheat by detecting whether the page is part of the test suite and running different code for it. |
| 17:20 | <jgraham> | In theory if you write good enough benchmarks that is true too |
| 17:20 | <Philip`> | AryehGregor: I think that was only reported on one SunSpider test, and it seems quite plausible that it's just hitting some optimisation threshold (don't inline functions larger than n AST nodes etc) or something |
| 17:20 | <AryehGregor> | Yes, if the benchmarks are real-world enough. |
| 17:20 | <jgraham> | I am unconvinced that sunspider is a good enopugh benchmark |
| 17:20 | <AryehGregor> | Philip`, maybe. |
| 17:20 | <AryehGregor> | No one seems to have any really good benchmarks. |
| 17:20 | <AryehGregor> | It's a shame. |
| 17:20 | <Philip`> | AryehGregor: I'd prefer more compelling evidence before accusing them of cheating on it |
| 17:20 | <jgraham> | It's a hard problem |
| 17:21 | <AryehGregor> | Philip`, yeah, okay. |
| 17:21 | <AryehGregor> | It's just an offhand remark on IRC. |
| 17:21 | <AryehGregor> | It was just an example, too. |
| 17:21 | <AryehGregor> | It happens all the time with this sort of thing. |
| 17:21 | <Philip`> | I agree it does happen |
| 17:22 | <AryehGregor> | Like I think Raymond Chen said that sometimes driver authors will ship drivers to Microsoft with an installer that sets a registry key that makes them act safely so that they pass the tests and get signed by Microsoft, then ship it to users with an installer that sets a different registry key so that it's faster but less stable. |
| 17:22 | <AryehGregor> | So totally different code paths. |
| 17:22 | <Philip`> | and there's lots of cases of not directly cheating but just fine-tuning the code specifically for the benchmarks at the expense of normal code |
| 17:22 | <AryehGregor> | It's very hard to detect that kind of thing in general with closed-source software. |
| 17:23 | <AryehGregor> | Actually, don't I remember hearing that some browser contained special-case code for a font used in a prominent test or something? |
| 17:23 | <AryehGregor> | (I think WebKit and Acid3, but my memory is very vague) |
| 17:23 | <Ms2ger> | You would be right |
| 17:23 | <Philip`> | Microsoft should just require than all programs running on Windows must be open source |
| 17:23 | <Philip`> | It'd solve a lot of their compatibility problems |
| 17:24 | <AryehGregor> | How could they enforce it without breaking compatibility with all existing programs? |
| 17:24 | <AryehGregor> | (hypothetically) |
| 17:24 | <Philip`> | Just break compatibility once and be done with it |
| 17:25 | <Ms2ger> | Versioning, of course! |
| 17:25 | <Philip`> | It's not like everyone's going to suddenly migrate to ReactOS |
| 17:25 | <AryehGregor> | No, but they'll migrate to pirated old Windows. |
| 17:26 | <Philip`> | Push out an automated update that kills all old Windowses |
| 17:26 | <Philip`> | Microsoft must know a lot of security vulnerabilities in old Windowses so they could write a pretty effective virus that'll catch even machines not on the internet |
| 17:27 | <AryehGregor> | Pirated Windows doesn't get automated updates. |
| 17:27 | <AryehGregor> | At least it's not supposed to. |
| 17:27 | <Philip`> | That's why you distribute the update in a virus |
| 17:28 | <AryehGregor> | That sounds illegal. |
| 17:28 | <AryehGregor> | They'd have to do it via a plausibly-deniable proxy. |
| 17:29 | <Philip`> | Disguise it as a virus that's infecting Iranian nuclear power stations or something |
| 17:35 | Philip` | just noticed http://hg.diveintohtml5.org/hgweb.cgi/rev/0c2514cfbd99 was committed "in the future ago" |
| 17:39 | <AryehGregor> | Weird . . . how easy is it to get a subdomain of mit.edu? http://margaret.mit.edu/ |
| 17:40 | <timdown> | Ms2ger: Hi. You seem to be editing a new version of the DOM Range spec. Being a random nobody, can I contribute? |
| 17:40 | <Ms2ger> | Yes! |
| 17:41 | <zcorpan> | timdown: awesome! |
| 17:41 | <Ms2ger> | timdown, do you have a bitbucket account? |
| 17:41 | <timdown> | Ms2ger: Yes, I signed up an hour or so ago |
| 17:42 | <timdown> | I should fork the repo, right? |
| 17:42 | <annevk> | should put a call for rand nobodies that are also aspiring spec editors on blog.whatwg.org :) |
| 17:42 | <annevk> | random, even |
| 17:42 | <Ms2ger> | timdown, I'll just give you access to the repo |
| 17:42 | <timdown> | great |
| 17:42 | Philip` | wonders if the Bitbucket acquisition is considered good or bad or unknown by open source people |
| 17:43 | <AryehGregor> | Unknown to me. |
| 17:44 | <Ms2ger> | And done |
| 17:45 | <Ms2ger> | timdown, you should be able to push |
| 17:45 | <Ms2ger> | if you need any help, do ping me |
| 17:47 | <zcorpan> | Ms2ger: innerText seems like a fun thing to spec |
| 17:48 | <Ms2ger> | zcorpan, I accept patches |
| 17:48 | <timdown> | Ms2ger, thanks. What's the aim with the new spec? Formalizing extensions? Ironing ambiguities in the old spec? |
| 17:48 | <Ms2ger> | Both |
| 17:48 | <zcorpan> | timdown: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10798 (fyi) |
| 17:49 | <timdown> | Ah, innerText being not dissimilar to how browsers implement Selection.toString()? |
| 17:49 | <Ms2ger> | Original goal was just speccing intersectsNode, tbh |
| 17:49 | <Ms2ger> | But that expanded a bit |
| 17:49 | <timdown> | zorcan: ah yes, I saw that. I wrote up a few Selection issues that prompted that, I think |
| 17:50 | <timdown> | intersectsNode works differently in Mozilla and WebKit |
| 17:50 | <timdown> | Mozilla took it out |
| 17:50 | <Ms2ger> | Yes |
| 17:50 | <Ms2ger> | And I have a bug assigned to me to add it back |
| 17:50 | <timdown> | heh, that was me as well |
| 17:50 | <annevk> | basically we want to rewrite the old DOM specs so they are far less ambiguous, match browsers more closely, and add/remove features based on existing implementations |
| 17:50 | <Ms2ger> | Oh, that's where I knew your name from |
| 17:50 | <timdown> | Seems likely |
| 17:51 | <jgraham> | annevk: You know it is shorter to say "don't suck"? |
| 17:52 | <Ms2ger> | jgornick, please rewrite that sentence so it doesn't suck |
| 17:52 | <Ms2ger> | s/jgornick/jgraham/ |
| 17:53 | <jgornick> | no i not rewrite better sentences |
| 17:53 | <jgornick> | :P |
| 17:53 | <Ms2ger> | Or rewrite your name not to interfere with tab-completing jgraham ;) |
| 17:53 | <timdown> | Ms2ger, just so I'm clear, you're happy for me to write up some of the stuff you haven't done yet on the Range spec, push it and then presumably there's some mechanism for picking it apart? |
| 17:54 | <timdown> | sorry |
| 17:54 | <timdown> | noob |
| 17:55 | <Ms2ger> | I would be very happy if you did that |
| 17:55 | <timdown> | OK |
| 18:11 | <AryehGregor> | while ! sudo -n true </dev/null >/dev/null 2>&1 ; do sleep 1; done ; sudo id </dev/null 2>&1 | cat |
| 18:11 | <AryehGregor> | Do that and then do something in the GUI that asks for your password, like updating the system. |
| 18:11 | <AryehGregor> | The command will immediately execute as soon as you type your password in the GUI. |
| 18:11 | <AryehGregor> | Yay for the requirement that you type your password actually doing anything. |
| 18:11 | <AryehGregor> | (this is assuming you're on Linux, obviously) |
| 18:26 | <karlcow> | AryehGregor: not tried, but it seems it could work on macosx too, or any bsd at least |
| 18:27 | <AryehGregor> | I wonder if there's an easy variant that works on Windows. |
| 18:32 | <Philip`> | I thought Windows didn't have the concept of global user elevation (like sudo apparently does), which is why it asks you a million times if you're sure you want to give permission to click this button and that button and that button |
| 18:34 | <AryehGregor> | It seems to be stricter than Linux, yes. |
| 18:34 | <AryehGregor> | But I don't know if it's actually asking you for every action. |
| 18:36 | <Philip`> | I think it's per-process, but there's often a lot of processes involved in a series of user actions (like setting up a new networking connection) |
| 18:36 | Philip` | has only used Vista, and assumes Win7 makes it somewhat better |
| 18:37 | <Hixie> | is there any way to make xml.dom.minidom.parse use less RAM? |
| 18:38 | <Hixie> | i'm trying to parse a 6MB file and it turns into hundreds of MB in RAM |
| 18:38 | <Philip`> | You could buy more RAM, then it will use proportionally less |
| 18:38 | <Hixie> | that's the solution i'm using |
| 18:38 | <Philip`> | Or you could use lxml, I guess |
| 18:38 | <Hixie> | but it's a pain to have to get more RAM each time i update the entities db |
| 18:38 | <AryehGregor> | You could also force it to swap everything, then it will use disk instead of RAM. |
| 18:39 | <Hixie> | i should say "memory" rather than RAM |
| 18:39 | <AryehGregor> | You're moving the goalposts. :( |
| 18:39 | <zcorpan> | you could use a custom regexp instead of a proper parser |
| 18:40 | <Hixie> | ok, having gotten more memory i can report that it uses 658MB |
| 18:40 | <Hixie> | i guess 100x increase isn't so bad (!) |
| 18:41 | <zcorpan> | or use a streaming parser maybe? |
| 18:41 | <tabatkins> | My browser tab doesn't have a 100x increase in memory usage when loading a page. |
| 18:41 | <Philip`> | Your browser's not written in Python |
| 18:41 | <Ms2ger> | You don't know that |
| 18:41 | <tabatkins> | This is true, but ideally it should be irrelevant. |
| 18:43 | <AryehGregor> | http://www.sharenator.com/w/whatwg.org |
| 18:43 | <annevk> | Hixie, the parsing does not happen elsewhere? |
| 18:43 | <Philip`> | Your browser was also written by people who spent a great deal of time optimising it, whereas Hixie's script was written by him presumably trying to do whatever's easiest to get working, using an API whose main design goal is to be reasonably simple |
| 18:46 | <Hixie> | annevk: elsewhere? |
| 18:48 | <annevk> | Hixie, I thought you generated it on pimpmyspec.net or some such |
| 18:50 | <jgraham> | Hixie: Just don't use minidom |
| 18:50 | <karlcow> | node.unlink() ? http://docs.python.org/library/xml.dom.minidom.html#dom-objects |
| 18:51 | <jgraham> | If you are using minidom you have already lost |
| 18:51 | <karlcow> | lxml is better than minidom for performance |
| 18:51 | <Hixie> | is there a drop-in replacement for minidom? |
| 18:51 | <Hixie> | annevk: nope |
| 18:53 | <jgraham> | Hixie: There are other dom modules, but not in the stdlib |
| 18:53 | <jgraham> | Hixie: Why are you so attached to DOM? |
| 18:53 | <Philip`> | It means one less API to learn |
| 18:54 | <tabatkins> | That's certainly why I use minidom for my page parsing needs. |
| 18:54 | <Philip`> | but lxml didn't seem much of a pain when I started using it |
| 18:54 | <jgraham> | Yeah, but a) DOm sucks and b) minidom sucks even more |
| 18:55 | <Hixie> | jgraham: because the code is already written and works fine, it just requires me to increase the memory allocation every few months when i have to rerun the entity processor script |
| 18:55 | <jgraham> | Hixie: How long is the code? |
| 18:55 | <Hixie> | and the pain of rewriting it exceeds the pain of 5 minutes of changing memory settings every few years |
| 18:55 | <Philip`> | Can't you run it on your local machine then upload the output? |
| 18:56 | <jgraham> | The pleasure of learning lxml so that you never have to use minidom again outweighs the cost of writing it |
| 18:56 | <Hixie> | the code is like 20 lines, the fixed cost of learning a new api is orders of magnitude higher than the incremental cost per line of changing the script :-) |
| 18:56 | <Hixie> | Philip`: far easier to briefly increase my memory allocation than try to ping pong data back and forth to my local dumb terminal |
| 18:57 | <Hixie> | jgraham: well to be fair i don't know how to use minidom either, i wrote this years ago |
| 18:57 | <Hixie> | and haven't used minidom since |
| 18:57 | <jgraham> | Oh well if it's 20 lines I will even rewite it for you. Though seriously, using minidom fr, well, anything, is a bad idea |
| 18:57 | <karlcow> | or publish the script, let someone fixing it |
| 18:58 | <Hixie> | jgraham: noted :-) |
| 18:59 | <Hixie> | http://damowmow.com/temp/entity-processor.txt is the script if anyone actually cares |
| 18:59 | <Hixie> | but it's fine, really |
| 18:59 | <Hixie> | i was just wondering if there was a quick fix i could apply |
| 19:01 | <Philip`> | You should compress your RAM |
| 19:05 | <AryehGregor> | Yes, but it would still use the same amount of virtual memory, and Hixie already ruled that out when I suggested swapping. |
| 19:05 | <AryehGregor> | Hixie, you have a local dumb terminal? |
| 19:06 | <tabatkins> | I didn't realize that Pokemon was actually set in the real Japan, until just now when I saw a map of all the pokemon areas overlaid onto the continent. |
| 19:06 | <tabatkins> | Now I'm mapping what the cities correspond to. |
| 19:06 | <Hixie> | AryehGregor: yeah, it's called a macbook pro |
| 19:09 | <AryehGregor> | Hixie, that's not a dumb terminal, it's a smart terminal. |
| 19:10 | <Philip`> | AryehGregor: Depends what you mean by "virtual memory" |
| 19:10 | <Hixie> | AryehGregor: i treat it as a dumb terminal |
| 19:10 | <AryehGregor> | tabatkins, seriously? Cool. |
| 19:10 | <Philip`> | The application would have the same virtual address space, but the OS could store it in fewer physical pages (either in RAM or as virtual memory on disk) |
| 19:11 | <AryehGregor> | Hixie, so you don't ever try to do things like resize your terminal or move the cursor backwards? |
| 19:11 | <tabatkins> | AryehGregor: http://i.imgur.com/bQg7E.png?ref=nf |
| 19:11 | <jgraham> | http://ietherpad.com/LuFCwnAf8s is something like what that might look like in lxml assuming the input has no namespaces, with added bugs, and without being any more idiomatic than the original |
| 19:11 | <AryehGregor> | tabatkins, I guess that would have been obvious to the original Japanese audience. |
| 19:12 | <Philip`> | Or I suppose you could set up a RAM disk, format it with a compressed filesystem, then put a swap file on it, so the OS doesn't have to do anything special |
| 19:12 | <tabatkins> | AryehGregor: Yeah, probably. The Kanto region is basically the area around Tokyo, so I assume that bit of geography is pretty well-known. |
| 19:13 | <tabatkins> | AryehGregor: Celadon City is very obviously Tokyo, for example. |
| 19:13 | <Ms2ger> | Oh, wow |
| 19:13 | <AryehGregor> | tabatkins, makes sense. |
| 19:13 | <AryehGregor> | Which cities did the bullet train go between? |
| 19:13 | <Hixie> | AryehGregor: resize my terminal, no, not other than by accident. It isn't a dumb terminal in the old mainframe sense, it's really just a dumb terminal in the Network Terminal sense Sun wanted -- it just runs a browser and some SSH sessions |
| 19:13 | Ms2ger | takes out his Game Boy |
| 19:13 | AryehGregor | only played Red/Blue and Gold/Silver |
| 19:13 | <tabatkins> | AryehGregor: That's from something other than red/blue/yellow, so I dunno. |
| 19:13 | AryehGregor | goes to dig up his Game Boy Color |
| 19:14 | <AryehGregor> | Oh, it is? |
| 19:14 | <AryehGregor> | I can't remember. |
| 19:14 | <tabatkins> | I'm playing Fire Red right now, so yes. |
| 19:14 | <AryehGregor> | It must be from GSC. |
| 19:14 | <AryehGregor> | Fire Red is what, a remake of Red with better graphics? |
| 19:14 | <AryehGregor> | I thought there was a train from Celadon City to somewhere. |
| 19:14 | <tabatkins> | Yeah. And an import of some of the 3rd gen mechanics. |
| 19:15 | <AryehGregor> | I guess I only know second gen. So like what mechanics? |
| 19:16 | <tabatkins> | Traits, and the dark/metal types. |
| 19:16 | <tabatkins> | Several new moves from the later games. |
| 19:17 | <AryehGregor> | Dark and steel are from generation two. |
| 19:17 | <AryehGregor> | Dunno about traits. |
| 19:17 | <tabatkins> | Right, but they exist in Gen 3 too. |
| 19:18 | <AryehGregor> | http://bulbapedia.bulbagarden.net/wiki/Magnet_Train |
| 19:18 | <AryehGregor> | It looks like it was between Goldrenrod and Saffron, and only in generators 2 and 3. |
| 19:18 | <tabatkins> | Ah, it's a Johto/Kanto link. |
| 19:18 | <AryehGregor> | Also: http://bulbapedia.bulbagarden.net/wiki/Pok%C3%A9mon_world_in_relation_to_the_real_world |
| 19:18 | <AryehGregor> | <3 wikis |
| 19:19 | <AryehGregor> | (also notice HTML5 versions, yay for 1.16.0) |
| 19:19 | <AryehGregor> | (also notice HTML5 features, yay for 1.16.0) |
| 19:19 | <AryehGregor> | (specifically <input type=search>) |
| 19:19 | <tabatkins> | Ah, I was wondering if Saffron was supposed to be part of Tokyo too. |
| 19:19 | karlcow | is suddenly missing Japan :) |
| 19:21 | AryehGregor | has never been to Japan and doesn't want to be there, but is suddenly missing Pokémon |
| 19:21 | karlcow | wonders where the original unicode.xml file is |
| 19:21 | AryehGregor | goes to find his GBC |
| 19:21 | <tabatkins> | Pokemon is a great game to take on long flights. It blows a lot of time, and is very kind to the battery. |
| 19:26 | <AryehGregor> | Turns out my Game Boy Color is right in the drawer, with Pokémon Gold loaded into it. Batteries still good and everything. |
| 19:26 | AryehGregor | checks where he was when he left off the game |
| 19:27 | <AryehGregor> | No game saved. :( |
| 19:28 | <tabatkins> | ;_; |
| 19:31 | <AryehGregor> | I've got Gold, Silver, Red, Blue, and Yellow here. Only Blue and Yellow have any saved games at all. Blue has a game by my sister, with one badge and 16 Pokédex entries. Yellow has a game by "Ash" with 0 badges and one Pokédex entry. |
| 19:32 | <AryehGregor> | I guess these things lose their memory if you don't use them for seven or eight years. :( |
| 19:32 | AryehGregor | got all the badges in both Red and Silver at some point |
| 19:32 | <tabatkins> | Right, they save games using an internal battery. |
| 19:32 | <tabatkins> | That probably means that the games themselves can no longer be relied on to save themselves. |
| 19:33 | <Hixie> | man, you try to paste one 80k URL into a browser, and that browser won't let you hear the end of it |
| 19:33 | Hixie | twiddles thumbs waiting for chrome to come back to him |
| 19:33 | <tabatkins> | This was before non-volatile flash memory was cheap. |
| 19:33 | <AryehGregor> | Hixie, you can't kill the tab? I guess the UI process is involved in URL parsing or something. |
| 19:33 | <AryehGregor> | tabatkins, really? Sad. I wonder why Blue and Yellow still work, then. |
| 19:33 | <AryehGregor> | I don't think they were used much more recently. |
| 19:33 | <tabatkins> | Luck, probably. |
| 19:33 | <Hixie> | firefox handled it much better |
| 19:34 | <tabatkins> | AryehGregor: A lot of NES games can't really be played anymore for the same reason, unless you open them up and replace the battery. |
| 19:34 | <AryehGregor> | :( |
| 19:34 | <tabatkins> | That's not an overly difficult task if you have a soldering iron. |
| 19:35 | <AryehGregor> | I have some SNES games lying around, including Super Mario RPG, which was my favorite. But I sold my SNES years ago for like $15, so they're kind of useless. |
| 19:35 | <tabatkins> | My friend's done it several times while cleaning the inside of used games he's bought. |
| 19:35 | <AryehGregor> | On the other hand, I can feel morally justified in playing a ROM if I own the actual game and there's no reasonable legal way to play it. |
| 19:35 | <tabatkins> | AryehGregor: Buy it on the Virtual Console! |
| 19:35 | <AryehGregor> | What's that? |
| 19:36 | <tabatkins> | Wii. |
| 19:36 | <AryehGregor> | Also, why do you change your name's capitalization sometimes? It changes coloring in XChat, which is distracting. |
| 19:36 | <AryehGregor> | Oh, really? |
| 19:36 | <AryehGregor> | But you have to buy it a second time, I suppose. |
| 19:36 | <TabAtkins> | Yeah, you do. But it's cheap, like $10 in wii points. |
| 19:37 | <TabAtkins> | And I don't change my name capitalization on purpose. lowercased seems to be the default I set up, so whenever I restart my comp it sets to that until I remember to identify myself and change it. |
| 19:37 | <AryehGregor> | "Wii points" meaning you have to buy them in blocks that don't cleanly divide the price of anything so you actually spend more than $10 and then have leftover points to try luring you into buying more things so that they aren't a waste? |
| 19:37 | <TabAtkins> | Actually, wii points generally divide well. |
| 19:37 | <TabAtkins> | Most prices are in blocks of $5. |
| 19:38 | <TabAtkins> | Otherwise, yes. |
| 19:38 | <TabAtkins> | There are some VC games that cost $6-$8, but shrug. |
| 19:48 | <Peter-> | Ms2ger: changed the post, thank you! |
| 19:49 | <Ms2ger> | np |
| 19:50 | <AryehGregor> | TabAtkins, regarding http://www.w3.org/Bugs/Public/show_bug.cgi?id=10647 : didn't you conclude that sigma = pixels/2 was the right formula, when we had that discussion on www-style? |
| 19:50 | <AryehGregor> | If so, yay. |
| 19:52 | <AryehGregor> | Also, now Philip` has to update all his canvas blur tests. |
| 19:56 | <TabAtkins> | AryehGregor: Yes. That's why that's the current definition in the CSS3 Backgrounds & Borders draft. |
| 19:56 | <AryehGregor> | Yay, convergence. |
| 19:57 | <AryehGregor> | Too bad SVG specifies sigmas. |
| 19:57 | <TabAtkins> | Yeah, it's a little annoying. But at least the rule is easy. When doing SVG, just cut your desired blur width in half. |
| 20:00 | <Hixie> | five points to the first person who can work out for me what the formal reference for EUC-JP should be |
| 20:00 | <Hixie> | IANA and Wikipedia have failed me |
| 20:08 | <AryehGregor> | annevk, does this fit your "send HEAD with a body" needs? http://httpd.apache.org/docs/2.0/mod/mod_asis.html |
| 20:08 | <AryehGregor> | (thanks to Julian for mentioning it in a bug I was reading) |
| 20:08 | <Philip`> | AryehGregor: Updating the tests should be trivial |
| 20:09 | Philip` | supposes he ought to just update it now, rather than sticking it on the end of an infinite queue |
| 20:12 | <karlcow> | Hixie: ISO 2022-JP (JIS) Text Codec |
| 20:15 | <AryehGregor> | "I'm filing this without testing what happens in other browsers and without providing bugzilla/cvs archeology information about the rationale of the lore, because I want to get this on file before the deadline." |
| 20:15 | <AryehGregor> | hsivonen, that sounds like cheating. |
| 20:15 | <AryehGregor> | Also, is it actually necessary? Anything that implementers agree on can still go in indefinitely, right? |
| 20:15 | <TabAtkins> | No, it's just an honestly incomplete bug. ^_^ |
| 20:16 | <karlcow> | ah no misread EUC-JP not JIS |
| 20:16 | <karlcow> | http://www.iana.org/assignments/character-sets |
| 20:17 | <karlcow> | Source: Standardized by OSF, UNIX International, and UNIX Systems |
| 20:17 | <karlcow> | Laboratories Pacific. Uses ISO 2022 rules to select |
| 20:18 | <AryehGregor> | Actually, I suppose the WHATWG cabal may as well totally ignore the HTMLWG deadline for bug-filing, since it presumably won't affect the WHATWG version regardless. |
| 20:19 | <Workshiva> | AryehGregor: Sort of like how HTMLWG ignored the WHATWG last call? :P |
| 20:19 | <AryehGregor> | ReportedBy: public-i18n-bidi⊙wo |
| 20:19 | <AryehGregor> | Bugs are being reported by mailing lists? |
| 20:19 | <Hixie> | karlcow: yes, read all those... but what's the reference? |
| 20:19 | <annevk> | AryehGregor, prolly not |
| 20:19 | <AryehGregor> | Workshiva, as far as I can tell, the WHATWG ignored the WHATWG last call too. Or at least I didn't see any difference other than the spec saying "Last Call" on it for a while. |
| 20:20 | <Hixie> | well shortly after we went to last call i switched to an unversioned model so it became kinda moot |
| 20:21 | <Philip`> | Hmph, the spec all got changed since I last updated the annotations :-( |
| 20:25 | <annevk> | AryehGregor, I should prolly draw a line too with how much of HTTP I test in the XMLHttpRequest test suite |
| 20:25 | <AryehGregor> | I've been trying to avoid testing anything outside of HTML5 in my reflection tests. |
| 20:25 | <AryehGregor> | Not WebIDL, DOM methods, etc. |
| 20:26 | <AryehGregor> | I do test that setAttribute() followed by getAttribute() returns the same thing (there are surprisingly many implementations where it doesn't always), but that's just to avoid confusing error messages when it causes reflection to fail since setAttribute() isn't working as intended. |
| 20:26 | <annevk> | do you try assigning null/undefined and see what happens? |
| 20:26 | <annevk> | that could really help Web IDL |
| 20:26 | <annevk> | that data |
| 20:27 | <annevk> | for things that take a DOMString that is |
| 20:27 | <AryehGregor> | No, I don't assign anything to the IDL attribute that's not in the range of the type, because that's WebIDL and I haven't read WebIDL, so I can't say what's supposed to happen. |
| 20:28 | <annevk> | I think your test should handle those boundaries |
| 20:28 | <annevk> | not quite sure what other test would |
| 20:31 | <Philip`> | Seems better to test too much than too little |
| 20:31 | <Philip`> | In the worst case, implementers will tell you your tests are wrong |
| 20:35 | <AryehGregor> | I could do it, but someone has to tell me the expected values. |
| 20:35 | <AryehGregor> | Or I have to read WebIDL. |
| 20:35 | <TabAtkins> | It's really simple for the null/undefined stuff. |
| 20:35 | <AryehGregor> | Maybe I just need to put this kind of thing off till when I have more free time. |
| 20:35 | <annevk> | Web IDL is quite easy to read |
| 20:36 | <annevk> | and soonish heycam will be able to help out |
| 20:36 | TabAtkins | became sufficiently expert to advise some dudes in about 5 minutes. |
| 20:36 | <annevk> | well, the null/undefined stuff is still not decided |
| 20:37 | <annevk> | i guess the defaults will be ""/"undefined" respectively, but it requires some more investigation |
| 20:37 | <TabAtkins> | Luckily my advice was regarding null/undefined passed to a numeric argument. |
| 20:37 | <AryehGregor> | I already sat down today thinking "Okay, let me do some schoolwork before I check my e-mail or anything." Then went "Eh, I did schoolwork all yesterday, plus the professor said he doesn't really care if we hand stuff in on time, let me quick check who pinged me on IRC." That was about five hours ago. |
| 20:38 | <AryehGregor> | Why aren't all values just cast to the type of the variable, using standard JS rules? Not web-compatible? |
| 20:38 | <annevk> | AryehGregor, heh, shut down IRC :) |
| 20:38 | <AryehGregor> | I'm capable of ignoring it. |
| 20:38 | <AryehGregor> | I just chose not to. |
| 20:38 | <annevk> | AryehGregor, not compatible no |
| 20:38 | AryehGregor | has a not-yet-installed copy of StarCraft II sitting a few feet away in case he gets really stressed |
| 20:39 | TabAtkins | is coming over to Aryeh's house. |
| 20:39 | <AryehGregor> | TabAtkins, I'm pretty sure you could get StarCraft II locally for less than the cost of a plane ticket. |
| 20:39 | <TabAtkins> | Yeah, but I can only run it on my wife's computer, and she's using it right now. |
| 20:39 | <AryehGregor> | Anyway, speaking of my reflection tests, Hixie kindly resolved a few months' worth of reflection-related bugs in one day, so I have some test-updating to do. |
| 20:40 | <AryehGregor> | TabAtkins, I suspect you get a computer capable of running StarCraft II locally for less than it would cost to get a last-minute plane ticket to New York. |
| 20:40 | <AryehGregor> | But I dunno, I don't fly much. |
| 20:41 | <TabAtkins> | I probably should avoid installing Steam on my work computer, though it would be *very* nice for playing Fallout on top settings. |
| 20:41 | <TabAtkins> | Mmm... |
| 20:42 | <AryehGregor> | Do you own your work computer? |
| 20:42 | <TabAtkins> | No. |
| 20:42 | <AryehGregor> | Google gave you a work computer with a decent graphics card? |
| 20:43 | <TabAtkins> | They gave me a work computer with a ridiculous everything. |
| 20:43 | <AryehGregor> | What's the graphics card? |
| 20:43 | Philip` | got given one with a decent graphics card, mainly because an indecent graphics card wouldn't have been able to power the 30" monitor |
| 20:43 | <karlcow> | aaaaaah |
| 20:43 | <AryehGregor> | I can understand they'd give programmers dual quad-core CPUs and 8 GB of RAM or whatever, but I can't see the need for a $300 graphics card. |
| 20:43 | <karlcow> | Hixie: I might have a lead |
| 20:43 | <karlcow> | ISO/IEC 10744 |
| 20:44 | <AryehGregor> | Philip`, are you working for someone these days? |
| 20:44 | <Philip`> | No |
| 20:44 | AryehGregor | isn't sure how much you'd need to spend to run Fallout 3 (I assume that's what you're referring to) on highest settings |
| 20:44 | <AryehGregor> | I mean, for a graphics card. |
| 20:44 | <AryehGregor> | My gaming machine has a . . . GeForce 7900, I think. |
| 20:45 | <TabAtkins> | nvidia quadro fx 380. No clue on the relative quality of that. |
| 20:45 | <AryehGregor> | I never cared much for running everything on highest settings, and it still works. |
| 20:45 | <AryehGregor> | http://www.nvidia.com/object/product_quadro_fx_380_us.html |
| 20:45 | <AryehGregor> | 256 MB of VRAM does not sound like a high-end card. Mid-end, maybe. But I don't follow this too closely. |
| 20:46 | <TabAtkins> | Eh, sure. It'd be fine to running F3 fairly high, though. |
| 20:46 | <AryehGregor> | Costs about $100. |
| 20:46 | karlcow | is grrrr at ISO spec |
| 20:55 | <annevk> | I'm not really sure it makes sense to do innerText in CSSOM |
| 20:55 | <annevk> | well, it may make sense, but I'd rather not take it on |
| 20:56 | <annevk> | for now |
| 20:56 | <karlcow> | http://en.wikipedia.org/wiki/JIS_X_0208 |
| 20:56 | <annevk> | and it's prolly better grouped with visual selection, editing, etc. |
| 20:56 | <karlcow> | "The official title of the current standard is 7-bit and 8-bit double byte coded KANJI sets for information interchange" |
| 21:34 | <jgraham> | "WebIDL is pretty easy" - which WebIDL is that? The one I have seen is quite confusing |
| 21:36 | <jgraham> | (it could just be me of course, but since other people seemed to find it non-trivial to understand and since they do things like program high performance javascript engines, I conclude that they are reasonably bright) |
| 21:37 | <jgraham> | (and hence that it is probably not just me) |
| 21:37 | <annevk> | http://www.icanhasmotivation.com/wp-content/uploads/2009/05/thebible.jpg heh |
| 21:37 | <TabAtkins> | The one I get from googling WebIDL. |
| 21:40 | <annevk> | jgraham, or maybe they have trouble reading specs? |
| 21:41 | <AryehGregor> | I found WebIDL confusing too, FWIW. |
| 21:41 | <jgraham> | antti_s: Possible, but unlikely |
| 21:41 | <annevk> | developers seem relatively often confused |
| 21:41 | <jgraham> | er |
| 21:41 | <AryehGregor> | Then again, some people find HTML5 confusing, and I think it's very clear. |
| 21:41 | <jgraham> | annevk: ^ |
| 21:41 | <AryehGregor> | (usually) |
| 21:41 | <jgraham> | Indeed, I think HTML5 is clear |
| 21:41 | <jgraham> | I think WebIDl is much less clear |
| 21:43 | <jgraham> | TabAtkins: FWIW the one you get by googling webidl (you are supposed to tell us not to verbify "Google") is the /TR/ version |
| 21:43 | <jgraham> | Which is hopelessly out of date |
| 21:43 | <TabAtkins> | I get the dev.w3.org version. |
| 21:45 | <jgraham> | TabAtkins: Obviously google biases good results towards its own employees |
| 21:45 | <jgraham> | Or Americans |
| 21:45 | <jgraham> | Or something |
| 21:45 | <TabAtkins> | Both, actually. |
| 21:45 | <TabAtkins> | That's your daily dose of confidential google-fu. |
| 21:47 | <TabAtkins> | I have to stop pulling up the asteroids-in-the-page bookmarklet and just spinning it in the center of the page spraying bullets every 5 minutes. |
| 21:49 | <annevk> | you get the dev version? impressive |
| 21:50 | annevk | just remembers URLs these days since Google fails hard |
| 21:50 | <AryehGregor> | Search results are customized based on your search history, you know. |
| 21:50 | <AryehGregor> | Try it in private browsing mode or whatever, perhaps tunneled over a server so it doesn't use IP address. |
| 21:51 | <jgraham> | I should try to convince google that w3.org/TR/ is all spam |
| 21:51 | <jgraham> | so it gets deindexed |
| 21:51 | <TabAtkins> | Yeah, a private mode window gives me /TR/ for the top result. |
| 21:51 | <annevk> | private gives me dev.w3.org as top |
| 21:51 | <annevk> | see, Google doesn't make any sense |
| 21:52 | <jgraham> | So, we need to get /TR/ deindexed |
| 21:52 | <jgraham> | Or destroy /TR/ |
| 21:52 | <jgraham> | Or both |
| 21:52 | <TabAtkins> | I go with "destroy /TR/". |
| 21:52 | <annevk> | pretty easy to just remember dev.w3.org/2006/webapi/WebIDL/ |
| 21:52 | <TabAtkins> | "easy", yeah. |
| 21:53 | <TabAtkins> | I can barely remember that css stuff is at /csswg/ |
| 21:53 | <jgraham> | annevk: Not really. Also not that handy if you tell someone else to look and they look in the wrong version |
| 21:53 | <jgraham> | Which happens *all the time* (as well you know) |
| 21:54 | <annevk> | so bad at sarcasm |
| 21:54 | <jgraham> | Oh |
| 21:54 | <jgraham> | The internet is bad at sarcasm |
| 21:54 | <jgraham> | Or rather text is bad at sarcasm |
| 21:54 | <annevk> | reddit is pretty good at it |
| 21:55 | <annevk> | at least after a couple of voting rounds :) |
| 22:05 | <karlcow> | "EUC-JP is based on the character repertoire defined in JIS X 0208, see http://tinyurl.com/2v8r9rp . It seems that the term "EUC-JP" is not defined within a standard, but is based on the encoding method first developed by AT&T and others in 1985." --- Felix Sasaki |
| 22:06 | <karlcow> | Hixie, ping |
| 22:06 | <AryehGregor> | You know what someone needs to do? Instrument their shipping browsers to store a list of all features -- like all supported elements, DOM attributes, JS features, maybe other stuff too -- and keep a persistent bitmap someplace that sets the bit for a feature when it's used. Then once in a while, when checking for updates or something, ship the bitmap back to the implementer. |
| 22:07 | <AryehGregor> | Maybe along with the number of pages the browser has visited since the bitmap was last cleared. |
| 22:07 | <AryehGregor> | Then you can say things like "99.99% of our users have not visited a page using feature X, so it's safe to remove." |
| 22:10 | <hober> | what webidl module are all of the IDL definitions in the spec supposed to be in? |
| 22:12 | <annevk> | if someone just /msg'd me I hit a wrong button so please repeat... |
| 22:14 | <AryehGregor> | While I'm at it, I also want browsers to be able to automatically sync passwords and cookies between them. |
| 22:14 | <AryehGregor> | And a pony. |
| 22:26 | <AryehGregor> | Can anyone tell me what I should put in my hgrc for committing to dvcs.w3.org, given that I don't generally use Mercurial and don't have any hgrc right now at all? I need to set ui.username to even record a commit, but what should I set so that "hg push" does what I want? |
| 22:26 | <AryehGregor> | IIRC, hg only supports HTTP auth, so I can't use a private key to authenticate, right? |
| 22:29 | <AryehGregor> | Looks like hg theoretically supports key-based login, but I guess password is easier to set up. |
| 22:33 | <jgraham> | [paths] |
| 22:33 | <jgraham> | default = https://dvcs.w3.org/html/ |
| 22:34 | <jgraham> | I think that is all you need |
| 22:34 | <AryehGregor> | Ah, it will just prompt me then. |
| 22:35 | <jgraham> | Yes |
| 22:36 | AryehGregor | just got an exception when running "hg pull", nice |
| 22:36 | <AryehGregor> | I'm impressed already. |
| 22:36 | <AryehGregor> | Also "hg push". |
| 22:37 | <AryehGregor> | Apparently it was my [auth] that did it. |
| 22:38 | <AryehGregor> | $ hg push |
| 22:38 | <AryehGregor> | pushing to https://dvcs.w3.org/hg/html |
| 22:38 | <AryehGregor> | searching for changes |
| 22:38 | <AryehGregor> | http authorization required |
| 22:38 | <AryehGregor> | realm: W3C Mercurial Repository |
| 22:38 | <AryehGregor> | user: agregor |
| 22:38 | <AryehGregor> | password: |
| 22:38 | <AryehGregor> | abort: HTTP Error 500: Internal Server Error |
| 22:38 | <AryehGregor> | Sigh. |
| 22:38 | <Philip`> | I think you get that if your password is wrong |
| 22:38 | <AryehGregor> | How helpful. |
| 22:39 | <jgraham> | I think that is a W3C problem though |
| 22:39 | <AryehGregor> | I'm not an HTTP expert, but I seem to recall that there might be a status code for that condition? Like maybe 403? (Or there might even be a more specific one for failed HTTP auth.) |
| 22:39 | <AryehGregor> | 500 is not helpful. |
| 22:39 | <jgraham> | rather than a mercurial problem per-se |
| 22:40 | <AryehGregor> | What login am I using, though? The same one I use to get access to the members' area? |
| 22:40 | <Philip`> | Yes |
| 22:40 | <Philip`> | like for polls and whatnot |
| 22:40 | <AryehGregor> | Chrome doesn't let me copy-paste it, maybe I mistyped it. |
| 22:40 | <AryehGregor> | jgraham, it's Mercurial's problem because Mercurial made the braindead decision to use HTTP auth, which is a vestigial authentication mechanism that nobody sane uses, cares about, or works on making non-broken. |
| 22:41 | <jgraham> | (agreed that 5000 is not helpful) |
| 22:41 | <AryehGregor> | If they used something normal like SSH, or their own custom protocol, the problem wouldn't occur. |
| 22:41 | <AryehGregor> | This seriously annoyed me when I tried to write a web app that integrated with hg. |
| 22:41 | <AryehGregor> | Changes to some auth information required restarting the web server. |
| 22:42 | <AryehGregor> | Or no, it was something about virtual paths, or I don't remember, but it was directly because hg only supports HTTP auth. |
| 22:42 | <AryehGregor> | </rant> |
| 22:42 | TabAtkins | still doesn't really understand HTTP Auth, and thus hates it. |
| 22:42 | <TabAtkins> | Also it's ugly. |
| 22:43 | <jgraham> | AryehGregor: Huh? ssh works fine with hg |
| 22:43 | <AryehGregor> | Does it? |
| 22:43 | <jgraham> | Yes |
| 22:43 | <AryehGregor> | This was a while ago. |
| 22:43 | <jgraham> | It has worked for at least two years |
| 22:43 | <AryehGregor> | Oh, well, I guess SSH was impractical for my purposes. |
| 22:43 | <AryehGregor> | Because I'd have had to give everyone SSH logins. |
| 22:43 | Philip` | wonders if AryehGregor is going to successfully push soon |
| 22:43 | <Philip`> | (because if so then I won't commit/push myself) |
| 22:43 | <Philip`> | (to save the effort of merging) |
| 22:44 | <AryehGregor> | Philip`, I just unsuccessfully pushed four times in a row, does that help? |
| 22:44 | <jgraham> | It is not clear to me that making shit up is better than reusing http auth |
| 22:44 | <AryehGregor> | Possibly. |
| 22:44 | <AryehGregor> | I remember being annoyed at the time, though. |
| 22:44 | <AryehGregor> | Anyway. |
| 22:44 | <AryehGregor> | I still can't commit. |
| 22:44 | <AryehGregor> | Probably it's just that I'm mistyping the password or something. |
| 22:44 | <jgraham> | Yeah |
| 22:45 | <AryehGregor> | Philip`, in git, you'd rebase before committing rather than having to merge, so it would be basically effortless if there were no conflicts. |
| 22:45 | <AryehGregor> | (which there wouldn't be, unless you're creating an AryehGregor directory in submission/) |
| 22:45 | <AryehGregor> | I take it hg doesn't support rebase out of the box. It seems to have lots of extensions. |
| 22:45 | <AryehGregor> | Extensions are bad. |
| 22:46 | <Philip`> | Does rebase alter the history? |
| 22:46 | <AryehGregor> | Maybe I'll get used to hg if I'm forced to use it enough, though. |
| 22:46 | <Philip`> | Hg seems to try hard to avoid ever altering history |
| 22:46 | <AryehGregor> | Yes, it rewrites all the local commits so they're based on the new upstream. |
| 22:46 | <AryehGregor> | It's extremely convenient. |
| 22:46 | <Philip`> | (which is kind of annoying and forces you to put a layer like mq on top of it) |
| 22:47 | <AryehGregor> | Although of course it completely breaks everything if you publish a tree and then rewrite it. |
| 22:47 | <AryehGregor> | Well, not completely, but it's a pain. |
| 22:47 | <AryehGregor> | But being able to freely rewrite local commits is one of the awesomest things about git. |
| 22:47 | <jgraham> | AryehGregor: With or without rebasing it will be rather painless if there are no conflicts |
| 22:47 | <AryehGregor> | Merge will have to record an extra commit, though, won't it? |
| 22:48 | <jgraham> | Well yes, but that is hardly painful |
| 22:48 | <jgraham> | W3C has an "email me my password in cleartext" facility if you need it BTW |
| 22:48 | <AryehGregor> | I've got it now. |
| 22:49 | <jgraham> | OK, good |
| 22:49 | <AryehGregor> | Extra useless commits are annoying, though. You want to avoid them. There's no reason to avoid a rebase. |
| 22:49 | <AryehGregor> | I mean, I haven't committed yet. |
| 22:49 | <AryehGregor> | But I'm sure what my password is. |
| 22:50 | <AryehGregor> | I give up. |
| 22:50 | <AryehGregor> | I can log in successfully to the members' area. |
| 22:50 | <AryehGregor> | But the same login for the DVCS fails. |
| 22:51 | <jgraham> | I assume you have the right username |
| 22:51 | <AryehGregor> | It works for the surveys. |
| 22:53 | <jgraham> | Talk to plh or MikeSmith? |
| 22:53 | <Philip`> | Huh |
| 22:53 | <Philip`> | abort: HTTP Error 500: Internal Server Error |
| 22:53 | <Philip`> | It's not just you |
| 22:53 | <AryehGregor> | Yay. |
| 22:53 | AryehGregor | commiserates |
| 22:53 | <Philip`> | (and I haven't changed my configuration since it last worked successfully) |
| 22:54 | <AryehGregor> | That's what they always say. |
| 22:55 | <jgraham> | Hmm |
| 22:55 | <jgraham> | I wonder if they changed the auth |
| 22:59 | <jgraham> | Yeah, I can't push either |
| 23:02 | <Philip`> | http://blogs.msdn.com/b/ie/archive/2010/09/29/ie9-s-faster-more-capable-compatibility-view-list.aspx |
| 23:02 | <Philip`> | <domain docMode="EmulateIE9" featureSwitch="createElementWithMarkup">verizonwireless.com</domain> |
| 23:02 | <Philip`> | Excellent - now there's not merely a dozen different engine modes, you can toggle individual features within each mode |
| 23:03 | <AryehGregor> | If that's what it takes to get IE to be standards-compliant by default, I'm sold. |
| 23:03 | <AryehGregor> | I don't think they had any other option than all this compatibility given how radically different IE was to every other browser. |
| 23:04 | <AryehGregor> | "We want the CV List to be as small as possible when IE9 ships and to get smaller over time because more and more of the web works well in IE9 by default." See, they don't want this to be permanent. |
| 23:04 | <AryehGregor> | It's a transition move. |
| 23:04 | <AryehGregor> | IE10 might not need it at all. |
| 23:05 | <Philip`> | "We won’t create feature switches for every change in behavior and sites cannot opt-in to them." |
| 23:05 | <Philip`> | Lack of opt-in is good |
| 23:06 | <AryehGregor> | So it's basically just what every browser does, except they use XML. |
| 23:06 | <Philip`> | so it seems more like Opera's browser.js |
| 23:06 | <AryehGregor> | And have more sites on the list. |
| 23:06 | <AryehGregor> | WebKit also has code like that, I know. |
| 23:06 | <AryehGregor> | Probably Gecko too. |
| 23:06 | <Philip`> | Yeah |
| 23:06 | <AryehGregor> | (One of WebKit's hacks is for old MediaWiki, yay.) |
| 23:19 | <Hixie> | even the spec has per-site stuff like that |
| 23:24 | <Hixie> | annevk: you don't like x-vendor-feature? |
| 23:24 | <Hixie> | annevk: it's the only one apple seems happy to use :-( |
| 23:40 | <Hixie> | i wonder if we should change the parser so that each node on the stack has an associated state, and when you pop the stack, you automatically switch back to the state that you were in when the node was pushed on |
| 23:42 | <Hixie> | the problem is HTML in foreign lands, where you really want to not be in foreign lands, but you need to be able to switch back when the first HTML-in-foreign element is popped |
| 23:42 | <Hixie> | same as the in-table-cell mode actually |
| 23:42 | <Hixie> | in-table-cell and in-foreign-content are basically the same problem |
| 23:43 | <Hixie> | in caption is like that too |
| 23:44 | <Hixie> | bbiab |
