| 00:59 | <G0k> | check it out all http://thefridgeowl.com/stuff/domeventtest/test01.html |
| 00:59 | <G0k> | server side event unit test thing |
| 01:00 | <G0k> | anyone with a recent copy of opera? |
| 01:04 | <Philip`> | "Recent" like 9.22, or like 9.5? |
| 01:04 | <G0k> | like..last 6 months i guess |
| 01:05 | <Philip`> | "Statement on line 50: Type mismatch (usually a non-object value used where an object is required)" |
| 01:05 | <G0k> | uh. of which file? |
| 01:06 | <Philip`> | (in 9.22, which is almost the most recent public one) |
| 01:06 | <Philip`> | Line 50 of linked script http://thefridgeowl.com/stuff/domeventtest/test01.js |
| 01:06 | <Philip`> | listener.addEventSource("test01.php"); |
| 01:06 | <G0k> | so do any of the Waiting fields turn to "Passed" ? |
| 01:06 | <colione> | three |
| 01:06 | <colione> | mouseover MouseEvent Passed |
| 01:06 | <colione> | mousemove MouseEvent Passed |
| 01:06 | <colione> | mouseout MouseEvent Passed |
| 01:07 | <Philip`> | They all say "Waiting" for me |
| 01:07 | <Philip`> | Oh, now those three say "Passed" |
| 01:07 | <colione> | Opera 9.22.3687 Mac Os X |
| 01:08 | <G0k> | oh heh |
| 01:08 | <Philip`> | They seem to go to "Passed" when I move my mouse over a bit just off the top-right corner of the table |
| 01:08 | <G0k> | because when you move your mouse over the thing |
| 01:08 | <G0k> | yeah |
| 01:08 | <Philip`> | Oh, I mean top rather than top-right |
| 01:08 | <G0k> | hm. so they don't seem to support the addEventListener method |
| 01:09 | <G0k> | lemme try the event-source tag |
| 01:10 | <colione> | same result in latest webkit |
| 01:10 | <G0k> | yeah i haven't checked my webkit implementation in yet. :) |
| 01:24 | <G0k> | wow well |
| 01:24 | <G0k> | http://thefridgeowl.com/stuff/domeventtest/test02.html |
| 01:24 | <G0k> | it does do a great job making opera leak memory anyway |
| 01:31 | <G0k> | finally got opera working...seems to dislike the divx plugin a lot |
| 01:31 | <G0k> | well if anyone wants to play with it http://thefridgeowl.com/stuff/domeventtest/test02.html now kinda works on opera 9.2.whatever |
| 01:32 | <G0k> | some things work some don't |
| 01:32 | <G0k> | dunno who's at fault for the nonworking ones |
| 01:32 | <Philip`> | I get some combination of "Passed", "Waiting" and "Not a (...)Event" on testf02.html |
| 01:33 | <Philip`> | Uh, test02.html |
| 01:33 | <G0k> | well. they seem to support some and not support others |
| 01:33 | <G0k> | and/or my tests are bad |
| 01:35 | <Philip`> | and/or the spec is bad |
| 01:36 | <G0k> | well that's almost undeniable. :) |
| 01:37 | <G0k> | my tests are particularly biased, since i kinda invented them to test my webkit implementation |
| 01:38 | <Philip`> | Opera's implementation is fairly old and untested, so I wouldn't expect it to be a good approximation of correctness. I've got no idea if they've made any improvements to this stuff for v9.5, though |
| 01:39 | <G0k> | is there a 9.5 public beta? |
| 01:39 | <Philip`> | Not yet |
| 01:39 | <Philip`> | The only date I've heard is "soon", for the past few months :-) |
| 01:40 | <G0k> | hm. i know there was a patch for an implementation in mozilla at some point too |
| 01:40 | <G0k> | but it was never accepted |
| 01:40 | <G0k> | kinda curious how that was implemented too |
| 01:42 | <colione> | i know that zcorpan_ has a copy of opera 9.5 *tries to draw his attention to irc* |
| 01:42 | <Philip`> | https://bugzilla.mozilla.org/show_bug.cgi?id=338583 |
| 01:43 | <zcorpan_> | someone want me to test something in kestrel? :) |
| 01:44 | <G0k> | http://thefridgeowl.com/stuff/domeventtest/test02.html |
| 01:44 | <colione> | :D |
| 01:44 | <Philip`> | zcorpan_: You could save a lot of effort if you accidentally sent an email to the WHATWG list with the Kestrel installer attached |
| 01:44 | <zcorpan_> | 10 passed, 7 failed, the rest are waiting |
| 01:44 | <zcorpan_> | Philip`: ha |
| 01:45 | <G0k> | k so doesn't look like any progress from 9.2 |
| 01:45 | <G0k> | zcorpan_: what about http://thefridgeowl.com/stuff/domeventtest/test01.html ? |
| 01:45 | <colione> | hmm 11 past in 9.22 |
| 01:45 | <colione> | passed* |
| 01:45 | <G0k> | heh uhoh |
| 01:46 | <G0k> | to be honest, my mechanism for type checking an event is really pathetic |
| 01:46 | <Philip`> | In what way? |
| 01:46 | <G0k> | i check to see if "event" is a XXXEvent by seeing if event.initXXXEvent is defined |
| 01:46 | <Philip`> | Ah |
| 01:46 | <zcorpan_> | G0k: 9 passed, 7 failed |
| 01:47 | <G0k> | zcorpan_: for test01? |
| 01:47 | <zcorpan_> | G0k: yes |
| 01:47 | <G0k> | ah neato so the addEventSource thing is in kestrel. thanks. |
| 01:47 | <Philip`> | Maybe you could do something like Object.prototype.toString(event) and see if it's called "[object XXXEvent]" |
| 01:50 | <G0k> | ah |
| 01:51 | <G0k> | alright well all those that were previously failing in opera now aren't |
| 01:52 | <G0k> | but they're not getting called either |
| 01:52 | <G0k> | i mean |
| 01:52 | <G0k> | the ones that weren't getting called |
| 01:52 | <G0k> | still are not |
| 01:54 | <Philip`> | "if (Object.prototype.toString(event) == interfaceName) { passed = false; ..." - that doesn't make sense |
| 01:55 | <Philip`> | It should be "if (... != "[object "+interfaceName+"]") ...", I think |
| 01:55 | <G0k> | well. it's returning just the class name |
| 01:55 | <G0k> | in both webkit and opera |
| 01:56 | <Philip`> | But you're setting passed to false if the value is equal to the expected value |
| 01:56 | <Philip`> | when it should be the other way around |
| 01:56 | <G0k> | oh craps |
| 01:56 | <G0k> | wow |
| 01:57 | <Philip`> | Also, I was wrong anyway |
| 01:57 | <Philip`> | It should be Object.prototype.toString.call(event) |
| 01:57 | <Philip`> | else you'll always get "[object Object]" |
| 01:57 | <Philip`> | Also this assumes the objects have the same type as the interface, rather than a subclass, and I don't know if that's valid |
| 02:00 | <G0k> | well i think the server side event spec currently explicitly says which class the events need to be |
| 02:01 | <G0k> | but yeah i mean i suppose it would still be valid if it were a subclass of that class |
| 02:03 | <G0k> | man javascript is really in need of "does this implement some interface" function |
| 02:06 | <zcorpan_> | foo.constructor ? |
| 02:07 | <G0k> | well for instance |
| 02:08 | <G0k> | Opera seems to use the KeyEvent prototype everywhere i would expect a KeyboardEvent prototype |
| 02:08 | <zcorpan_> | hm |
| 02:08 | <G0k> | which...i now notice isn't in dom 3. heh |
| 02:09 | <zcorpan_> | anyway... well beond bedtime for me |
| 05:30 | <Hixie> | "Niether Hixe nor Philip looked for use of <input usemap> . Hixie looked for misuse (or esoteric use) of <input usemap>." |
| 05:30 | <Hixie> | i thought i had specifically looked for _valid_ cases... |
| 05:30 | <Hixie> | i wonder what rob is basing that on |
| 05:31 | <G0k> | good god is this still being argued? |
| 05:32 | <G0k> | is there seriously a legitimate use for image maps at all nowadays? |
| 05:34 | <Hixie> | i dunno i'm not entering this discussion any more |
| 05:34 | <G0k> | deprecate with extreme prejudice! |
| 05:35 | <Lachy> | Hixie, Rob is looking for usage that matches the way he imagines <input usemap> should work, not the way it has ever been defined or implemented |
| 05:36 | <Lachy> | and thus, he rejects any observation of <input usemap> on the basis that what he's looking for automatically can't exist |
| 05:41 | <Hixie> | anyone tried explaining the whole thing about how scientific research and such is all about looking at what actually is there, and not what you're expecting? |
| 05:43 | <Lachy> | I think he misses the point of the whole observation stage. I think he wants to start directly with some hypothesis developed without any observation at all and then test it |
| 05:44 | <Lachy> | I don't think he understands that a hypothesis should be developed to try and explain your observations, when then gets tested and refined |
| 05:47 | <Lachy> | maybe someone needs to explain that the statistics he's objecting to so much are used as a rough guide only. The more important information is obtained from looking at how and why something gets used, which is what our decisions are based on |
| 05:47 | Hixie | shrugs |
| 05:49 | <G0k> | maybe it should be decided by a duel. |
| 05:50 | <Hixie> | on a more positive note, i do like how much the various studies correlate |
| 05:51 | <G0k> | anyone know who operates http://validator.nu/ ? |
| 05:55 | <Hixie> | hsivonen |
| 05:55 | karlUshi | would prefer that people do not use "scientific research" to justify, specifically when things are not really done that way. :) Or at least that there are many ways of doing scientific research. |
| 05:56 | karlUshi | has the feeling that people are mixing experimental research with scientific research. :) |
| 05:58 | <Lachy> | karlUshi, observation and experementation is part of the whole scientific process, so I don't understand what you mean |
| 05:59 | <karlUshi> | Lachy: My whole academic background is… scientific research :))) |
| 05:59 | <karlUshi> | Astrophysics exactly |
| 05:59 | <karlUshi> | there are many ways of doing sciences, not only one. |
| 06:00 | <karlUshi> | You can start from observations. You can also start from a model and looking for evidences by observing. |
| 06:00 | <karlUshi> | Observations don't necessary come first in the scientific method. |
| 06:12 | <G0k> | yeah but...once you fail to find evidence, that's not evidence of a conspiracy to undermine your model. |
| 06:39 | <Hixie> | i hope the hixie.ch dns gets fixed soon so i don't lose too much e-mail |
| 06:42 | <othermaciej> | is Rob just engaging in a brilliantly subtle attempt to drain all productivity? |
| 06:42 | <othermaciej> | I'm starting to be amazed at his ability to disagree with even the most basic things |
| 06:48 | <G0k> | is this on public-html? |
| 06:52 | <othermaciej> | yes |
| 06:53 | <othermaciej> | public-html is now mostly useful content plus him plus replies to him |
| 06:53 | <G0k> | who is he? |
| 06:53 | <othermaciej> | Robert Burns? |
| 06:54 | <G0k> | yeah what's he do? |
| 06:54 | <othermaciej> | it's hard to explain without experiencing it |
| 06:54 | <G0k> | i mean does he have an occupation? |
| 06:54 | <othermaciej> | oh |
| 06:55 | <othermaciej> | I believe he is a grad student |
| 06:55 | <karlUshi> | http://www.robburns.com/ |
| 06:56 | <karlUshi> | I guess he will be able to come at the Tech Plenary being in Mass. |
| 06:56 | <G0k> | does he design pages or something? |
| 06:57 | <G0k> | eeeks econ say no more |
| 06:58 | <othermaciej> | marxist econ |
| 06:58 | <karlUshi> | G0k: this is an innapropriate comment, except if you let me say "Computer Programmer… oh these binary people! say no more.". :/ |
| 06:59 | <Lachy> | what is Marxian philosophy? |
| 06:59 | <G0k> | i make try to make as few appropriate comments as possible |
| 06:59 | <G0k> | it is my intention to maximize my offendees |
| 06:59 | <othermaciej> | presumably the philosophy of Karl Marx, his students, and intellectual followers |
| 06:59 | <G0k> | so yeah bring it on |
| 07:00 | <othermaciej> | I don't have a problem with economists as a class |
| 07:00 | <karlUshi> | othermaciej: me neither. |
| 07:00 | <karlUshi> | and computer programmer as well. |
| 07:00 | <othermaciej> | most of them are good at thinking in terms of incentives, spontaneous order, and unintended consequences |
| 07:00 | <karlUshi> | There are people of different nature and different style, specifically online/offline. |
| 07:00 | <othermaciej> | all of which are are useful concepts to apply to global technical standards |
| 07:01 | <karlUshi> | I know a few persons on this channel who behave differently online and offline. |
| 07:01 | <G0k> | i have several economist friends, they're just all very likely to get into an argument with me |
| 07:02 | <karlUshi> | maybe you have to ask questions about yourself then G0k ;) |
| 07:02 | <G0k> | yeah i don't deny that i'm likely to be wrong usually |
| 07:54 | <Lachy> | Hixie, would it be possible for the spec to provide some sort of heuristic guidelines for programmatically determining when should be provided? e.g. if you have <h1><img src=""></h1>, it should be clear for a validator that the lack of alt text is an error. |
| 07:55 | <othermaciej> | you could require alt in any context that would require text if the embedded content wasn't there |
| 07:55 | <Lachy> | could it be defined that alt="" can only ever be omitted when contained within a <figure>, since that would represent it as a key part of the content |
| 07:55 | <othermaciej> | or something |
| 07:56 | <Lachy> | currently, all the examples in that section where alt can be omitted are contained within figure too |
| 07:58 | <othermaciej> | allowing missing alt only in <figure> would address the photo gallery use case but not the "pasting images into an end-user rich text editor" use case |
| 07:59 | <Lachy> | hmm, yeah |
| 08:56 | <jgraham> | Lachy: I would be _very_ wary of trying to subset where alt="" is allowed. |
| 08:56 | <jgraham> | Oh wait a sec, I misread that |
| 08:56 | <jgraham> | Ignore me |
| 08:56 | <Lachy> | right |
| 09:13 | <virtuelv> | I really wish people could refrain from bad ad hominems on random blogs |
| 09:13 | <Lachy> | virtuelv, ? |
| 09:13 | <Lachy> | pointer |
| 09:18 | <virtuelv> | http://www.456bereastreet.com/archive/200708/the_html_5_circus_why_i_left_and_rejoined_the_w3c_html_working_group/#comment3 |
| 09:21 | <krijnh> | Yeah, thanks for DOSing my server Roger ;] |
| 09:24 | <krijnh> | Ah, the comment |
| 09:25 | <othermaciej> | looks like Google has secretly invented the fountain of youth |
| 09:27 | <virtuelv> | krijnh: yes, I was specifically refering to the comment itself |
| 09:28 | <virtuelv> | While I may not agree with all of what Roger says, he is at least not resorting to fallacies to argue his position |
| 09:30 | <othermaciej> | Roger's actual post seems reasonable, though looking back at his posts to the group and the replies to them, I'm not sure why he felt attacked |
| 09:30 | <Lachy> | "I am far from the only person to think that the current process of editing the specification for the next version of HTML is… less than ideal." - we really need to address this issue somehow. |
| 09:30 | <Lachy> | although the consensus based approach some people are advocating won't work, I think we need better PR to improve the current process |
| 09:31 | krijnh | agrees |
| 09:31 | <krijnh> | Although I don't know how to fix it |
| 09:31 | <hsivonen> | the IETF has *rough* consensus and *running* code |
| 09:32 | <hsivonen> | much better than consensus on theory |
| 09:32 | <krijnh> | Perhaps it's the whatwg "vs." w3c attitude |
| 09:32 | <othermaciej> | I like the idea of the IETF process |
| 09:32 | <othermaciej> | though I have no real experience with the practice |
| 09:32 | <Lachy> | Roger just suggested to me that, whenever Hixie sends one of his summary emails responding to an issue on whatwg, we should notify public-html of it too |
| 09:33 | <krijnh> | whatwg is easier to track though |
| 09:33 | <hsivonen> | Lachy: good idea |
| 09:34 | <krijnh> | Why isn't that done already? |
| 09:34 | <krijnh> | Seems logical to do so |
| 09:34 | <Lachy> | getting a schedule from Hixie too would be good, and searching the whatwg archives and providing pointers to previous discussions beforehand may also help |
| 09:36 | <othermaciej> | there's lots of organizational and informational things that could be done, but ideally by someone other than Hixie since his plate is pretty full |
| 09:36 | <othermaciej> | I feel like the sense that reported issues are tracked is not solid enough yet, which may make people feel ignored |
| 09:36 | <hsivonen> | Lachy: last time when Hixie had an open-secret schedule, it didn't hold. I think the best we could get is a topic list without calendar time |
| 09:36 | <Lachy> | yeah, I'm sure others will volunteer to find all that info. We just need to know what the upcoming topics are |
| 09:36 | <othermaciej> | a short list of the next few topics would be helpful |
| 09:37 | <othermaciej> | and could guide review as well |
| 09:37 | <othermaciej> | (more useful to review soon-to-be-revisited areas) |
| 10:10 | <virtuelv> | I'd like to have something like trac for open issues |
| 10:20 | <hsivonen> | hrm. looks like the Dreamhost DNS is down |
| 10:21 | <virtuelv> | hsivonen: appears that way, all my sites are down as well |
| 10:46 | <Lachy> | is the only purpose behind the suggested noalt="" attribute (or equivlent) just a hack to work around the validation issue? |
| 10:48 | <hsivonen> | Lachy: the purpose is to make hand coders perform an affirmative act of typing instead of just leaving something out |
| 10:49 | <hsivonen> | (fwiw, I don't like the noalt idea. It won't help with non-hand-coded source and it will majorly annoy hand coders) |
| 10:49 | <Lachy> | hmm. Ok. But I think it's perfectly fine for authoring tools and conformance checkers to provide warnings when an alt attribute has been omitted, if the author wants such information |
| 10:50 | <Lachy> | indeed, and it doesn't acctually address the issue of getting people to provide it, or just abusing it to pass validation |
| 10:50 | <hsivonen> | yeah, warnings requested by the author are very different from baking something into the conformance definition |
| 10:53 | <Lachy> | you summed it up nicely here http://www.mail-archive.com/whatwg⊙lwo/msg06146.html |
| 10:54 | <hsivonen> | btw, if you feel there are particular issues that I should develop optional warnings for, please start a page on the WHATWG wiki |
| 11:02 | <Lachy> | I think there is a page on the wiki for that stuff somewhere, isn't there? |
| 11:02 | <Lachy> | I'll have to check when the server is back up |
| 11:55 | <Whiskey_M> | 'lo |
| 12:52 | <gsnedders> | is it just me or is whatwg.org down? |
| 13:14 | <mgdm> | gsnedders: I can't resolve it |
| 13:14 | <mgdm> | in DNS, I mean |
| 13:14 | <Lachy> | all of Hixie's sites are down. I think it's a problem with dreamhost |
| 13:14 | <mgdm> | Ah, I tried to get to simplebits.com earlier and it was broken too |
| 13:20 | <Philip`> | dreamhoststatus.com appears to be broken too, which doesn't help immensely |
| 14:28 | <gsnedders> | what's the point of the mask in "Content-Type sniffing: unknown type"? |
| 14:29 | <gsnedders> | ah. to make them case insensitive, or not. |
| 16:28 | <Lachy> | would anyone like to review this blog post for the whatwg blog (which I'll publish when the site is back up) http://lachy.id.au/temp/alt |
| 16:40 | <zcorpan_> | Lachy: looks good to me... though you may want to include more examples of where it is ok to omit alt |
| 16:41 | <hsivonen> | Lachy: I'd do s/the reality of the situation/the situation/ to avoid the impression that others are missing the reality |
| 16:41 | <Lachy> | I'm going to be linking to the spec, should I just borrow some of those? |
| 16:41 | <Lachy> | I like talking about reality, but ok |
| 16:42 | <Lachy> | I suppose it hasn't gone down well with some people |
| 16:43 | <jonbarnett> | my lynx doesn't behave as described here |
| 16:44 | <Lachy> | it doesn't? There's emails in the whatwg archive that say it does |
| 16:45 | <zcorpan_> | jonbarnett: you have a test url? |
| 16:45 | <jonbarnett> | for me, lynx has options to ignore images, treat images as labels, or treat them as links. When treated as labels, the filename is used when alt is omitted, and nothing is used when alt is blank. When treated as links, the filename is used both when alt is omitted and when alt is blank. It's not that big of a difference: lynx is still making a distinction, and your version may vary |
| 16:47 | <Lachy> | I'm not sure what version was tested, it wasn't mentioned |
| 16:47 | <jonbarnett> | (I may have been the one mentioning it and not having tested thoroughly *blush*) |
| 17:01 | Philip` | wonders if it could be required for conformant <table>s to contain at least one <th> |
| 17:02 | <Philip`> | since I can't think of any obvious cases where you'd have a table with no headings, unless it's a layout table |
| 17:02 | <zcorpan_> | Philip`: i've used tables without headers that aren't layout tables |
| 17:03 | <zcorpan_> | genealogical table |
| 17:08 | <Lachy> | Philip`, the question you need to ask is, in the event that there are use cases for tables without headers, what harm is caused by the lack of th? |
| 17:09 | <hsivonen> | Philip`: four-cell tables in MBA material |
| 17:09 | <hsivonen> | Philip`: unless you consider all four to me <th>s |
| 17:09 | <hsivonen> | s/to me/to be/ |
| 17:18 | <Philip`> | Lachy: If there are uses cases for headerless data tables, then we can't require headers; but if there aren't, then conformance checkers could know that any headerless table is a non-data table and is therefore non-conforming |
| 17:18 | <Philip`> | (e.g. http://www.noraradcliffe.co.uk/ is (very nearly) conforming XHTML1, designed by people who "exceed accessibility requirements", but uses tables for layout) |
| 17:18 | <hsivonen> | Philip`: which would only lead to people including bogus or barely-semantic <th>s in their tables |
| 17:18 | <Lachy> | maybe, but it's very difficult to prove that there aren't any |
| 17:19 | <Philip`> | (Er, I mean it's (very nearly) valid XHTML1) |
| 17:23 | <Philip`> | http://www.w3.org/TR/xhtml-role/ - "A conforming user agent MUST support all of the features required in this specification." - hmm, I can't actually find any required UA features in the specification |
| 17:26 | <zcorpan_> | Philip`: indeed |
| 17:28 | <zcorpan_> | e.g. http://www.pointerklubben.se/stamtavla.asp?Id=S27834/2001 is a table with no headers |
| 17:42 | <hsivonen> | zcorpan_: interesting dog names |
| 17:58 | <doublec> | is www.whatwg.org down? |
| 17:59 | <Philip`> | Yes |
| 18:00 | <Philip`> | (Dreamhost's problem, I think) |
| 18:02 | <doublec> | thanks |
| 18:23 | <h3h> | hooray Dreamhost! |
| 18:23 | h3h | rolls eyes |
| 18:32 | <jonbarnett> | ... and hixie's site is down, too? |
| 18:34 | <Philip`> | Yes, for the same reason |
| 18:36 | <Hixie> | this is the worst dreamhost outage yet |
| 18:36 | <Hixie> | if it wasn't for the fact that i pay $0, i'd be outraged |
| 18:36 | <Hixie> | yesterday i was worried that i was losing whatwg e-mail |
| 18:37 | <Hixie> | but now that whatwg isn't resolving either, i'm less concerned :-P |
| 18:37 | <Philip`> | At least http://www.dreamhost.com/ is up and says "Unfortunately we are experiencing a return of similar network problems that we experienced yesterday. Please keep your eye on this post as we will update it as soon as we have mor..." then links to the dead dreamhoststatus.com |
| 18:37 | <Philip`> | so presumably they have actually noticed there's a problem |
| 18:37 | <jonbarnett> | my personal space is on DH, and it's up (I guess I'm just tauting) |
| 18:38 | <met_> | http://ajaxian.com/archives/web-forms-20-cross-browser-implementation |
| 18:41 | <zcorpan_> | hmm, some of the tests don't have any pass condition |
| 18:42 | <zcorpan_> | and all have too much junk around the actual test... :) |
| 18:47 | <Hixie> | they do know there's a problem |
| 18:47 | <Hixie> | they've been working overnight trying to fix it |
| 18:47 | <Hixie> | they update their internal status message ever few minutes |
| 18:48 | <Hixie> | something to do with lost dns records, they're rebuilding their dns config from scratch using a slow-working script |
| 18:48 | <Hixie> | the discussion in #html-wg was interesting |
| 18:49 | <Hixie> | can someone summarise in an unbiased way what robburns' concerns were? |
| 18:51 | <Hixie> | Lachy: btw i apparently have ~3800 or so (+/- 200, i'm going from memory) e-mails in my queue right now |
| 18:52 | <Lachy> | oh, nice! It's reduced :-) |
| 18:52 | <zcorpan_> | aiui, he wanted the html wg to discuss issues raised outside the group before the spec is edited, instead of the editors editing based on external feedback and then letting the wg discuss further... but i might have missed something |
| 18:52 | <jonbarnett> | Hixie: i haven't been on #html-wg. is it about <img> or your editing process? |
| 18:53 | <jonbarnett> | nm |
| 18:55 | <hsivonen> | Hixie: my understanding was that Rob was unhappy that 1) you are focusing on stuff other than the discussions of public-html and 2) that your recent edits (in his opinion) run counter the thrust of discussions on public-html. (this may well be a biasod account) |
| 18:56 | <Hixie> | ok |
| 18:57 | <Hixie> | 2) may well be true, i've seen no real consensus on public-html so it's hard to know what the thrust of discussion is |
| 18:57 | <Hixie> | i guess i could start picking htmlwg issues to deal with if it makes them happier |
| 18:57 | <Lachy> | Hixie, I believe Rob would agree with these suggestions http://krijnhoetmer.nl/irc-logs/whatwg/20070817#l-320aren't |
| 18:57 | <Hixie> | doesn't really make much difference what order i deal with issues in |
| 18:57 | <hsivonen> | I've been concerned about the formation of a false consensus on public-html |
| 18:57 | <Lachy> | http://krijnhoetmer.nl/irc-logs/whatwg/20070817#l-320 |
| 18:58 | <hsivonen> | Rob's view suggests that my concern is valid |
| 18:58 | <hsivonen> | Hixie: well, it does make a PR difference |
| 18:58 | <Hixie> | i can certainly bcc public-html on any mail to whatwg |
| 18:58 | <hsivonen> | Hixie: that would probably help |
| 18:59 | <Hixie> | the reason i haven't is that it would cause a flood of e-mail to whatwg |
| 18:59 | <Hixie> | since every e-mail i send to whatwg is just replied by people with actual experience, but e-mail sent to public-html immediately devolves into a flamewar |
| 19:00 | <hsivonen> | but it's really tiresome to avoid the formation of false consensus by keeping parroting old points |
| 19:00 | <Hixie> | don't worry about stopping false consensus |
| 19:00 | <hsivonen> | Hixie: true |
| 19:00 | <Lachy> | I think the alt attribute thread has actually become somewhat productive now |
| 19:01 | <Hixie> | just make sure that the wiki page for the issue clearly states all points |
| 19:01 | <Lachy> | although, started out as another near-flameware |
| 19:01 | <Hixie> | Lachy: really? you mean i should make more changes beyond what the spec says? |
| 19:01 | <Lachy> | Steven Faulkner wrote a constructive suggestion. I'm not sure I agree with it, but it was mostly well written |
| 19:03 | <Hixie> | you mean http://lists.w3.org/Archives/Public/public-html/2007Aug/0660.html ? |
| 19:03 | <Hixie> | i don't see how that could work |
| 19:03 | <Lachy> | nor do I. I was just pointing out that it was well intended, not a flame |
| 19:04 | <Hixie> | (btw, re your idea, <h1></h1> isn't non-conforming -- in fact we're dropping the concept of significant inline content) |
| 19:04 | <Hixie> | sure |
| 19:04 | <Lachy> | hmm. that's what I thought you'd say |
| 19:05 | <Hixie> | it's a good idea, though, i'd do it if we weren't doing the other thing |
| 19:05 | <Hixie> | is there a wiki page with a list of problems that have been fully described and are stable? |
| 19:06 | <Lachy> | I suppose if conformance checkers implement a feature like that, you could spec it. |
| 19:07 | <Lachy> | in the HTMLWG wiki? I don't think anything is particularly stable there |
| 19:08 | <Hixie> | maybe i should send mail saying that i'm still lwaiting for a stable wiki page |
| 19:09 | <Lachy> | the table headers issues is probably the most stable, and addressing that sooner rather than later would probably really help calm people down |
| 19:11 | <Lachy> | good night everyone. |
| 19:12 | <Hixie> | nn |
| 19:13 | <Hixie> | hmm |
| 19:14 | <Hixie> | i'm gonna go to work and consider how else to help keep public-html in the loop |
| 19:15 | <Hixie> | i don't think forwarding all whatwg mail to public-html (or even my own e-mails to public-html) would be productive |
| 19:15 | <Hixie> | but maybe we should try that anyway |
| 19:16 | <Hixie> | hmm |
| 19:16 | <Hixie> | afk |
| 19:17 | <takkaria> | I think that just having a public list with the next few things you're thinking of working on, with some semi-accurate dates, would go a long way |
| 20:08 | <Hixie> | takkaria: yeah... the problem is right now what i'm working on is "pick random piece of feedback, reply to it" |
| 20:09 | <takkaria> | may I suggest such a list might also be useful to you, then? :) |
| 20:09 | <Hixie> | woot, my site is back up :-D |
| 20:09 | <Philip`> | That sounds like it could be extended to "pick several random pieces of feedback, put them in a list, and reply to them in that order" |
| 20:09 | <Hixie> | takkaria: i dunno, i find i work most productively on the opposite of what i'm supposed to work on :-) |
| 20:10 | <takkaria> | ah, so you're not superhuman either? shame |
| 20:10 | <Hixie> | takkaria: literally, to the point where i've sometimes made lists of what i Must Do Now that are actually the opposite of what i really have to do to trick myself into doing it :-P |
| 20:10 | <takkaria> | heh |
| 20:11 | <takkaria> | I can't imagine a list of things you're not going to work on being very useful to the WG, since that'd be quite a long list |
| 20:11 | gsnedders | can think of one think to do: |
| 20:11 | gsnedders | faceplams |
| 20:11 | <gsnedders> | *facepalms |
| 20:11 | <gsnedders> | Hixie: the worst part is I'm not actually that surprised at you doing that. |
| 20:12 | <Hixie> | heh |
| 20:12 | <Hixie> | so, anyone want to take bets on what lists i got unsubscribed from due to my mail bouncing? |
| 20:13 | <gsnedders> | I'm tempted to guess public-xhtml |
| 20:14 | <Philip`> | Mailman seems to default to needing several weeks of bouncing before unsubscribing someone |
| 20:14 | <Hixie> | cool |
| 20:15 | <Philip`> | (but I could be wrong, or people could have non-default settings) |
| 21:19 | <kingryan> | does anyone know if anyone has implemented "4.7.4. Content-Type sniffing: feed or HTML" |
| 21:19 | <kingryan> | ? |
| 21:20 | <kingryan> | sorry, my client is acting up |
| 21:20 | <gsnedders> | kingryan: yes. me. |
| 21:20 | <kingryan> | gsnedders: do you have a test suite? ;) |
| 21:20 | <gsnedders> | kingryan: no |
| 21:20 | <gsnedders> | kingryan: the implementation is literally just finished :) |
| 21:20 | <kingryan> | btw, url: http://www.whatwg.org/specs/web-apps/current-work/#content-type3 |
| 21:21 | <kingryan> | gsnedders: is your code available? |
| 21:21 | <kingryan> | open source? |
| 21:21 | zcorpan_ | could perhaps write some tests on content sniffing |
| 21:21 | <gsnedders> | zcorpan_: I'll write some tonight |
| 21:21 | <kingryan> | I'll contribute some tests too |
| 21:21 | <zcorpan_> | gsnedders: ok, great :) |
| 21:22 | <gsnedders> | kingryan: it will be, soon |
| 21:22 | <kingryan> | I have an interpretation question though |
| 21:22 | <gsnedders> | BTW, for step 10: http://lists.w3.org/Archives/Public/public-html/2007Aug/0671.html |
| 21:22 | <kingryan> | in step 6 of that algorithm |
| 21:22 | <gsnedders> | the double otherwise? |
| 21:22 | <kingryan> | both substeps 3 and 4 start with "otherwise" |
| 21:22 | <kingryan> | :) yes |
| 21:22 | <gsnedders> | a mistake, without question |
| 21:22 | <gsnedders> | for the algorithm to make sense they need to be a single "otherwise" |
| 21:22 | <kingryan> | did you get a clarification from Hixie or whoever else contributed that? |
| 21:23 | <zcorpan_> | how did you implement it? if ... else ... else ? :) |
| 21:23 | <gsnedders> | not yet, but there's no other way the algorithm would even work |
| 21:23 | <gsnedders> | (<http://lists.w3.org/Archives/Public/public-html/2007Aug/0672.html>;) |
| 21:23 | <kingryan> | ok, I'll try that approach |
| 21:24 | <gsnedders> | how should test cases be done? |
| 21:24 | <gsnedders> | as the HTTP headers are relevant… |
| 21:24 | <gsnedders> | just entire HTTP responses? JSON objects of header name => value? |
| 21:25 | <zcorpan_> | hmm |
| 21:25 | <kingryan> | I know 4.7.4 isn't self contained, but I don't see any references in that sectoin to http |
| 21:25 | <gsnedders> | kingryan: but section 4.7 as a whole |
| 21:25 | <zcorpan_> | gsnedders: just using .htaccess should work, no? |
| 21:25 | <kingryan> | right. I'm only focused on 4.7.4 right now |
| 21:26 | <kingryan> | but for testing the section as a whole... |
| 21:26 | <gsnedders> | zcorpan_: I currently have all my tests in SVN though, and not served through Apache |
| 21:26 | <kingryan> | what I usually do is have pairs of files "foo.html" and "foo.json" |
| 21:27 | <kingryan> | s/html/http/ |
| 21:27 | <kingryan> | where "foo.http" is a full http response (headers and all) |
| 21:27 | <kingryan> | and "foo.json" includes whatever data we hope to extract |
| 21:27 | Philip` | wonders if that content-type sniffing algorithm can be correctly implemented as a regular expression |
| 21:27 | <kingryan> | Philip`: I think most state machines can be mapped to regexen |
| 21:28 | <gsnedders> | Philip`: depends on the regular expression engine :P |
| 21:29 | <Philip`> | I guess it'd be somewhat like ... |
| 21:29 | <Philip`> | my $ct = 'text/html'; $ct = { rss => 'application/rss+xml', feed => 'application/atom+xml' }->{$1} if substr($data, 0, 512) =~ /^(?:\s+(?:<!--.*?-->|!.*?>|\?.*?\?>))+?(rss|feed)/; |
| 21:29 | <gsnedders> | kingryan: so .json would just have the MIME type, in this case? |
| 21:29 | <Philip`> | at least if you ignore the rdf:RDF bit |
| 21:29 | <kingryan> | gsnedders: yes |
| 21:29 | <Philip`> | but that's quite utterly untested |
| 21:29 | <gsnedders> | Philip`: see the former of the two emails I linked to about rdf:RDF |
| 21:30 | <gsnedders> | Saf just treats <rdf:RDF as application/rss+xml, and I'm unaware of anything that breaks due to it |
| 21:31 | <Philip`> | I guess /^(?:\s+(?:<!--.*?-->|!.*?>|\?.*?\?>))+?<(rss|feed|rdf:RDF)/ might do that, then |
| 21:32 | <Philip`> | Urgh, but I'm still missing some <s |
| 21:32 | <Philip`> | /^(?:\s+(?:<!--.*?-->|<!.*?>|<\?.*?\?>))+?<(rss|feed|rdf:RDF)/ |
| 21:32 | <Philip`> | /^(?:\s+(?:<!--.*?-->|<!.*?>|<\?.*?\?>))+?<(rss|feed|rdf:RDF)/ |
| 21:33 | Philip` | finds that much easier to read than what's in the spec |
| 21:33 | <Philip`> | though it's still wrong |
| 21:33 | <Philip`> | /^(?:\s*(?:<!--.*?-->|<!.*?>|<\?.*?\?>))*?<(rss|feed|rdf:RDF)/ |
| 21:34 | <Philip`> | /^(?:\s*(?:<!--.*?-->|<!.*?>|<\?.*?\?>))*?<(rss|feed|rdf:RDF|)/ |
| 21:34 | Philip` | will get there eventually |
| 21:36 | <Philip`> | Hmm, still wrong |
| 21:36 | <Philip`> | /^(?:\s*(?:<!--.*?-->|<!.*?>|<\?.*?\?>))*?<(rss|feed|rdf:RDF|(?!!--|!|?))/ |
| 21:37 | <Philip`> | Okay, maybe it's getting less easy to read |
| 21:40 | <gsnedders> | how about /^<\?xml[\x20\x9\xD\xA]+version([\x20\x9\xD\xA]+)?=([\x20\x9\xD\xA]+)?(\"1.0\"|'1.0'|\"1.1\"|'1.1')[\x20\x9\xD\xA]+encoding([\x20\x9\xD\xA]+)?=([\x20\x9\xD\xA]+)?(\"[A-Za-z][A-Za-z0-9._\-]*\"|'[A-Za-z][A-Za-z0-9._\-]*')([\x20\x9\xD\xA]+standalone([\x20\x9\xD\xA]+)?=([\x20\x9\xD\xA]+)?(\"(yes|no)\"|'(yes|no)'))?([\x20\x9\xD\xA]+)?\?>/ |
| 21:40 | <gsnedders> | :P |
| 21:41 | <gsnedders> | if that's easy to read… |
| 21:43 | <Philip`> | It'd be a bit nicer if you split the common whitespace bits into a variable and did /^<\?xml $ws+ version $ws* = .../x or something :-) |
| 21:44 | <gsnedders> | it's a good question why I put it all together before saving it. I wrote it with variables like that. |
| 21:44 | <Philip`> | I wonder if I can compile the tokeniser algorithm into a Perl regexp... |
| 21:44 | <gsnedders> | Philip`: try! I doubt it, but I want to see the result if you can |
| 21:44 | <gsnedders> | kingryan: actually, just storing whole HTTP responses gets annoying when you have binary responses |
| 21:46 | <Philip`> | I don't remember seeing any bits of the algorithm that would cause annoying pumping lemma problems when trying to make a (theoretical) regular expression of it, and in any case nobody cares about that since real regexps are not regular... |
| 21:47 | <kingryan> | gsnedders: why is that annoying? |
| 21:47 | <takkaria> | being able to define your own character classes could increase readability a lot |
| 21:47 | <Philip`> | (I don't think a great number of theoretical models of regular expressions even attempt to cope with the ability to include arbitrary Perl code in the middle of one) |
| 21:48 | <gsnedders> | kingryan: text editors tend to not like arbitrary binary data |
| 21:48 | <kingryan> | right |
| 21:49 | <gsnedders> | kingryan: you want to do all of feed/html? |
| 21:50 | <kingryan> | gsnedders: not sure what you're asking |
| 21:50 | <gsnedders> | kingryan: of the test cases |
| 21:50 | <gsnedders> | kingryan: if you do the feed/html ones, I'll just do the rest |
| 21:50 | <kingryan> | cool |
| 21:51 | <kingryan> | yeah, I'll do as many as it takes for my software to work right :) |
| 21:51 | <gsnedders> | the feed/html ones are actually the ones I need, too :P |
| 21:52 | <kingryan> | so, how do we want to organize these? |
| 21:52 | <gsnedders> | that's what I'm wondering. |
| 21:52 | <kingryan> | the feed/html ones aren't dependent on HTTP |
| 21:52 | <kingryan> | is "content-type" the only HTTP header we need? |
| 21:53 | <gsnedders> | there's one check that content-encoding isn't set |
| 21:53 | <gsnedders> | but that only applies for certain content-type headers anyway |
| 21:53 | <gsnedders> | (step 1, 4.7) |
| 21:54 | <kingryan> | in other words, do we need arbitrary http headers? |
| 21:54 | <kingryan> | if not, we can just use JSON |
| 21:55 | <takkaria> | surely you could use JSON anyway? |
| 21:55 | <kingryan> | takkaria: the question was whether we need full http headers |
| 21:55 | <kingryan> | which I doubt we do |
| 21:55 | <gsnedders> | kingryan: we could just have tests for sections 4.7.[1-4] |
| 21:55 | <kingryan> | gsnedders: it'd be useful for me to have tests for the subsections, but we should probably also do overall tests |
| 21:55 | <gsnedders> | kingryan: actually no |
| 21:56 | <kingryan> | no what? |
| 21:56 | <gsnedders> | kingryan: the image section relies on the outer algorithm if it doesn't find anything |
| 21:56 | <kingryan> | ah, ok |
| 21:56 | <kingryan> | nevermind then |
| 21:56 | <gsnedders> | unknown jumps to text or binary |
| 21:57 | <kingryan> | we can still separate them logically so that if you have only implemented, for example "html vs. feed", you can get that section of tests to pass |
| 21:57 | <gsnedders> | yeah |
| 21:57 | <kingryan> | how about we structure the tests like so: |
| 21:57 | <kingryan> | [{"input": "512 byte string", "result": "mime type"},...] |
| 21:58 | <gsnedders> | always 512 bytes? |
| 21:58 | <kingryan> | isn't that the limit? |
| 21:58 | <kingryan> | unless I'm mistaken |
| 21:58 | <gsnedders> | "512 or more" |
| 21:58 | <gsnedders> | it's the min |
| 21:59 | <kingryan> | it seems to me like there non-determinism there |
| 21:59 | <gsnedders> | we better keep them to 512 bytes, so the most pessimistic implementations pass |
| 22:00 | <kingryan> | that seems reasonable to me |
| 22:00 | <gsnedders> | what do you want to do for the "Big Issue"? |
| 22:00 | <kingryan> | nothing for now? |
| 22:00 | <kingryan> | we have plenty of non-"big issue" things to implement |
| 22:01 | <gsnedders> | that's true. |
| 22:02 | <gsnedders> | kingryan: why did you use [{…}] though? |
| 22:02 | <gsnedders> | why the outer array? |
| 22:02 | <gsnedders> | you mean to have them all in one file? |
| 22:02 | <kingryan> | I was just saying that we'd have a list of them in json |
| 22:03 | <kingryan> | yes, I mean to have more than one per file |
| 22:03 | <gsnedders> | right. |
| 22:03 | <kingryan> | and we should do a file for each sub-section |
| 22:04 | <kingryan> | so binary.json, unknown.json, image.json, feed.json |
| 22:06 | <gsnedders> | JSON has no sort of hex encoding, only unicode, doesn't it? |
| 22:06 | <kingryan> | I think so |
| 22:06 | <gsnedders> | doesn't that make it impossible to represent some of the bytes we need? |
| 22:07 | <kingryan> | like what? |
| 22:08 | kingryan | has only really looked at the feeds sectoin |
| 22:08 | gsnedders | realises. |
| 22:08 | <gsnedders> | I can just do \u00FE\u00FF |
| 22:08 | <gsnedders> | allowing a UTF-16BE BOM without using the unicode code point for it. |
| 22:08 | <kingryan> | gsnedders: we could also do arrays of numbers |
| 22:08 | <gsnedders> | kingryan: no, strings are easier |
| 22:08 | <kingryan> | I agree |
| 22:09 | <gsnedders> | kingryan: do you have commit access to put them in html5lib? |
| 22:09 | <kingryan> | yes |
| 22:09 | <gsnedders> | ah. I don't. |
| 22:10 | <kingryan> | I can commit tests if you send them to me |
| 22:10 | gsnedders | instinctively types \xFF and not \u00FF |
| 22:11 | <gsnedders> | kingryan: @technorati.com one? |
| 22:11 | <kingryan> | either that or @theryanking.com |
| 22:11 | <kingryan> | doesn't matter |
| 22:11 | <gsnedders> | I just looked at what I had on mailing lists from you :) |
| 22:12 | <gsnedders> | jgraham: around? |
| 22:12 | <jgraham> | Yep |
| 22:13 | <gsnedders> | jgraham: possible to add me to html5lib so I can add tests? |
| 22:14 | <jgraham> | Sure. foolistbar? or some other address? |
| 22:14 | <gsnedders> | geoffers |
| 22:14 | jgraham | needs a better gmail address |
| 22:15 | <gsnedders> | foolistbar sometimes seems so appropriate |
| 22:15 | <jgraham> | OK it should be done |
| 22:15 | Philip` | needs a better one too |
| 22:16 | <gsnedders> | jgraham: thanks |
| 22:16 | <Philip`> | If someone could kick off all the early users who got the good usernames, that'd be very helpful |
| 22:16 | <gsnedders> | like me? |
| 22:16 | gsnedders | ducks |
| 22:16 | <gsnedders> | I got foolistbar quite late on, actually |
| 22:16 | <gsnedders> | they were already using googlemail.com by then |
| 22:16 | <jgraham> | gsnedders: You don't have anything like as common a name as Philip` or I though |
| 22:16 | <gsnedders> | jgraham: that's true |
| 22:18 | jgraham | decides jgrham.html might be good for html related bits |
| 22:18 | <jgraham> | er jgraham that is |
| 22:19 | <jgraham> | or if not good at least better |
| 22:23 | <gsnedders> | god do I hate typing JSON |
| 22:23 | <gsnedders> | practice makes perfect, though |
| 22:25 | <gsnedders> | (all of the tests I've so far written read "Hello, World!") |
| 22:25 | <takkaria> | I like using Jabberwocky when string data is required for tests |
| 22:26 | <jgraham> | Has anyone implemented the table model stuff |
| 22:26 | <jgraham> | ? |
| 22:27 | jgraham | can't see where the algorithm is supposed to move on to the next row |
| 22:30 | <gsnedders> | http://geoffers.no-ip.com/svn/php-html-5-direct/tests/type-sniffing/textOrBinary — that's horrid to work on :\ |
| 22:31 | <kingryan> | gsnedders: that does look horrid |
| 22:32 | <kingryan> | working with binary data in a text editor is never fun though |
| 22:32 | <gsnedders> | those aren't binary. |
| 22:32 | <gsnedders> | they're specific unicode encodings. |
| 22:32 | <kingryan> | ah, ok |
| 22:32 | <gsnedders> | BOM + "Hello, World!" |
| 22:33 | <zcorpan_> | don't \u... represent characters, not bytes? |
| 22:33 | <gsnedders> | zcorpan_: yes, but there's no other way to do it :( |
| 22:33 | <zcorpan_> | ok |
| 22:33 | <kingryan> | maybe JSON isn't the best approach then? |
| 22:33 | <gsnedders> | everything is going to be horrid |
| 22:34 | <zcorpan_> | [0xFE,0xFF,0x00, ... |
| 22:34 | <gsnedders> | that's the alternative in JSON |
| 22:34 | <kingryan> | gsnedders: we *could* do integer arrays |
| 22:34 | <gsnedders> | yeah |
| 22:34 | <gsnedders> | probably better, in actuality |
| 22:34 | <kingryan> | and that'd be closer to how the spec writes it |
| 22:35 | <kingryan> | how about we say the tests runners should be able to take either a string or an array of integers representing bytes? |
| 22:35 | <gsnedders> | actually, no hex forms in JSON. |
| 22:35 | <gsnedders> | ergh. |
| 22:35 | <kingryan> | ick |
| 22:36 | <gsnedders> | if we're doing JSON, the way it is is the only hex way. |
| 22:36 | <gsnedders> | yuk. |
| 22:37 | <Philip`> | Maybe you could have an array of string+ints, so you can write [0xFE,0xFF,"Hello",0x20,"world"] |
| 22:37 | <gsnedders> | Philip`: no. hex. numbers. |
| 22:37 | <Philip`> | Oops - [254,255,"Hello",32,"world"] |
| 22:38 | <gsnedders> | that means converting everything from the spec |
| 22:38 | <gsnedders> | and I can't convert hex to dec in my head |
| 22:38 | <gsnedders> | slowwww :( |
| 22:38 | <Philip`> | JSON5! |
| 22:38 | <gsnedders> | :) |
| 22:38 | <Philip`> | Or just say it's JS |
| 22:38 | <gsnedders> | yes, but half the point of JSON is it is supported in other languages |
| 22:39 | <Philip`> | Or use strings like "FE FF 00 48" and get people to translate it after the JSON parsing |
| 22:39 | <gsnedders> | XML + numeric entities suddenly looks tempting |
| 22:46 | <kingryan> | gsnedders: maybe a microformat? ;) |
| 22:46 | <gsnedders> | kingryan: what an out of the world suggestion from you :) |
| 22:47 | <kingryan> | :) |
| 22:47 | <gsnedders> | kingryan: <var> could be used for input and result… but their content? |
| 22:48 | <gsnedders> | <samp> for result? |
| 22:48 | <kingryan> | sure |
| 22:48 | <gsnedders> | and for the input? |
| 22:49 | <kingryan> | <code> ? |
| 22:50 | <gsnedders> | I'm slightly dubious about that, but I can't think of anything better |
| 22:50 | <kingryan> | <pre> ? |
| 22:50 | <gsnedders> | probably better we use XHTML, so there is no need for a full blown HTML parser |
| 22:51 | <gsnedders> | <pre> suits us insofar as that the entities (as they'll have to be) are the preformated input |
| 22:52 | <gsnedders> | actually, we can't have � |
| 22:52 | kingryan | likes the idea of requiring an HTML parser for the HTML test suite |
| 22:53 | <kingryan> | ok, so are we back to JSON, then? |
| 22:53 | <gsnedders> | what happens to � in HTML5? Hixie? |
| 22:53 | <gsnedders> | *� |
| 22:54 | <kingryan> | I don't see any references to that in html5 |
| 22:54 | <Philip`> | "if the number is zero, if the number is higher than 0x10FFFF, or if it's one of the surrogate characters (characters in the range 0xD800 to 0xDFFF), then this is a parse error; return a character token for the U+FFFD REPLACEMENT CHARACTER character instead" |
| 22:55 | <Philip`> | ( http://www.whatwg.org/specs/web-apps/current-work/multipage/section-tokenisation.html#tokenising ) |
| 22:55 | <kingryan> | ah |
| 22:55 | <gsnedders> | I knew if it was unescaped it got replaced |
| 22:55 | gsnedders | sighs |
| 22:56 | <gsnedders> | damn. null. bytes. |
| 22:56 | <Philip`> | You could store the whole document with ROT-1, so the null bytes map onto  |
| 22:57 | takkaria | chuckles |
| 22:57 | <kingryan> | I think we're back to JSON |
| 22:57 | <gsnedders> | ROT-1 sounds like a nice idea |
| 22:57 | <gsnedders> | as long as we never need 01 |
| 22:58 | <kingryan> | you mean FE? |
| 22:58 | <gsnedders> | it means we can't test one byte in a range of bytes |
| 22:58 | <gsnedders> | kingryan: no, with rot-1 |
| 22:58 | <gsnedders> | kingryan: map � to  |
| 22:58 | <kingryan> | that mens FE maps to 00, right? |
| 22:58 | <gsnedders> | kingryan: ah. true. |
| 22:59 | <gsnedders> | kingryan: FF, surely? |
| 22:59 | <Philip`> | You've got the whole of Unicode to work in - use Ā |
| 22:59 | <Philip`> | (for ÿ + 1) |
| 22:59 | <kingryan> | true |
| 22:59 | <gsnedders> | or just replace 00 with 01 |
| 22:59 | <gsnedders> | and keep the rest as is |
| 22:59 | <kingryan> | as long as we don't need FFFE :) |
| 22:59 | <gsnedders> | FFFE is UTF-16 BOM |
| 23:00 | <kingryan> | I know ;) |
| 23:00 | <gsnedders> | I expected you did |
| 23:00 | <gsnedders> | replace 0 with 1 seems the simple solution |
| 23:02 | <kingryan> | are we still talking about doing this in html? |
| 23:03 | <gsnedders> | yes. |
| 23:08 | <gsnedders> | RFC: http://geoffers.no-ip.com/svn/php-html-5-direct/tests/type-sniffing/textOrBinary |
| 23:09 | <gsnedders> | wow. we've taken 75 minutes to get this far. |
| 23:09 | tantek | volunteers gsnedders to go to a W3C f2f meeting. |
| 23:10 | <gsnedders> | tantek: if you can get me out of school and get me paid for, sure :P |
| 23:10 | gsnedders | wonders what is up with tantek's connection |
| 23:11 | <gsnedders> | tantek: if you can get me out of school and get me paid for, sure :P |
| 23:11 | <gsnedders> | tantek: what's up with your connection, BTW? |
| 23:12 | <kingryan> | I think tantek's at a cafe |
| 23:12 | <gsnedders> | and bad wi-fi? |
| 23:14 | <gsnedders> | kingryan: should we go with HTML then? |
| 23:14 | <gsnedders> | kingryan: just have it marked up like that? |
| 23:15 | <gsnedders> | tantek: why me? |
| 23:18 | <tantek_> | gsnedders, you mentioned something about taking so many minutes to get so far. indicating that you've clearly learned some patience. a very useful quality. |
| 23:19 | <gsnedders> | tantek_: not patience, just persistence |
| 23:19 | <tantek_> | another useful quality. |
| 23:20 | <jgraham> | Since test case formats are tonight's hot topic, would anyone like to suggest a format for table model tests? They need to take HTML as an input and check that the cells appear in the right places in a 2D grid |
| 23:20 | <jgraham> | JSON would work but it would be really verbose |
| 23:21 | <jgraham> | I guess some custom plaintext format might be best |
| 23:22 | <jgraham> | Oh and each slot can have >1 or 0 cells |
| 23:22 | <jgraham> | and each cell has extra metadata which needs to be checked :( |
| 23:24 | <Philip`> | You could represent the 2D grid with an HTML table |
| 23:26 | <kingryan> | gsnedders: as cool as using HTML would be, many of the tests are already in JSON and adding more dependencies, even if self-dependencies adds to the cost of collaboration |
| 23:26 | <kingryan> | jgraham: I'd use a multidimensional JSON array |
| 23:27 | <kingryan> | [[cell, cell], [cell, [double, cell]]] |
| 23:27 | <gsnedders> | kingryan: we'd need to add something like "Every unicode character must be interpreted as a byte with the value of the unicode code point the character represents" |
| 23:27 | <kingryan> | yes |
| 23:28 | <jgraham> | kingryan: That's sort of thing is basically what I was thinking too. It's just slightly irritating that json has such a rigid syntax. However it's less verbose than HTML |
| 23:29 | Philip` | likes YAML |
| 23:29 | <kingryan> | right, but json is always utf-8, so the code-point=>character=>byte relationship is fixed |
| 23:29 | <kingryan> | Philip`: json is yaml :) |
| 23:30 | <kingryan> | http://redhanded.hobix.com/inspect/yamlIsJson.html |
| 23:30 | kingryan | uses a yaml parser for json sometimes |
| 23:30 | <Philip`> | JSON is only a subset of YAML, so YAML necessarily has less of a rigid syntax :-) |
| 23:31 | <kingryan> | true |
| 23:31 | <kingryan> | and it does allow hex numbers |
| 23:32 | <Philip`> | YAML seems hugely more complex to parse than JSON, but I don't care because I just say "use YAML::Syck" and it works magically |
| 23:32 | <kingryan> | indeed |
| 23:32 | <kingryan> | I'd rather not add another dependency to html5lib, though |
| 23:32 | <kingryan> | (though that's a no-op for ruby, since yaml is in the stdlib) |
| 23:32 | <gsnedders> | kingryan: "JSON text SHALL be encoded in Unicode. The default encoding is UTF-8. " |
| 23:33 | kingryan | stands corrected |
| 23:33 | gsnedders | has RFC4627 open due to his sheer lack of knowledge of JSON and remembers noticing that earlier |
| 23:41 | <gsnedders> | kingryan: you go and make up some sort of test case format |
| 23:41 | <gsnedders> | time for me to sleep |
| 23:41 | <kingryan> | k |