| 02:45 | <heycam> | which is correct en-US: "acknowledgment" or "acknowledgement"? |
| 02:50 | <Hixie> | both |
| 02:50 | <heycam> | :/ |
| 02:51 | <heycam> | well, the one with the "e" has more hits on google, so i'll go with that |
| 02:51 | <heycam> | using google hit counts is the ultimate descriptivist approach to language =) |
| 02:59 | <roc> | I can't wait for Dictionary5 which abandons normative spelling rules in favour of documenting existing usage so other language users don't have to reverse engineer it |
| 03:00 | <Hixie> | we have normative spelling rules? |
| 03:00 | <heycam> | imagine that, a defined way to parse ambiguous sentences |
| 03:00 | <Hixie> | the only normative spec i know of for english is ian.hixie.ch/bible/english |
| 03:01 | <Dashiva> | Hixie: I'm sure your English teacher had one, just not written down :) |
| 03:01 | <roc> | if we don't have normative spelling rules, then my school teachers lied to me! |
| 03:01 | <Hixie> | roc: many do :-) |
| 03:11 | <hallvors> | Hixie: you've probably written some tests for script execution/scheduling, right? |
| 04:30 | <Hixie> | hallvors: nothing useful as far as i recall |
| 04:33 | <hallvors> | Hixie: I've written some 60+ tests, need to push a couple of buttons to get them out .. |
| 04:34 | <Hixie> | cool |
| 04:34 | <hallvors> | they are not very well tested against other browsers and some may of course assert things that will be wrong according to the spec |
| 04:35 | <hallvors> | but I plan to get feedback from other vendors and you when they are out |
| 04:37 | <hallvors> | hope you'll find them useful :) |
| 08:54 | <jgraham> | http://www.foms-workshop.org/foms2009/pmwiki.php/Main/TimedText |
| 08:57 | hsivonen | wonders if anyone is publishing non-bogus hard drive MTBF numbers by hard disk brand |
| 09:07 | <annevk> | ok, what am I missing? |
| 09:07 | <annevk> | <div><button></div>X |
| 09:07 | <annevk> | in Opera and Firefox the "X" is not inside the <button> |
| 09:07 | <annevk> | afaict in HTML5 <button> is closed at </div> |
| 09:08 | <annevk> | both html5lib and livedom.validator.nu put X inside the <button> |
| 09:08 | <Hixie> | <button> is scoping in html5 |
| 09:08 | <Hixie> | so it isn't closed at </div> |
| 09:09 | <hsivonen> | which reminds me that livedom is very much out of date |
| 09:12 | <annevk> | ah, you're right |
| 09:12 | annevk | misremembered the definition of "has an element in scope" |
| 09:13 | <hsivonen> | livedom updated (reload caches, etc.) |
| 09:13 | annevk | wonders which browsers does this like HTML5 then |
| 09:13 | <Hixie> | probably firefox, though maybe not with <div> |
| 09:14 | <Hixie> | no browser exactly matches the parser algorithm |
| 09:14 | <annevk> | IE6 does it |
| 09:17 | <annevk> | Opera 10 does it |
| 09:27 | <annevk> | http://shouldiusetablesforlayout.com/ :p |
| 09:29 | <Lachy> | LOL. Check the comments in the source of that page |
| 09:29 | <annevk> | I know, hence the :p |
| 10:01 | <annevk> | what version of Safari has querySelector support? |
| 10:02 | annevk | summons olliej |
| 10:05 | <Lachy> | annevk, Safari 4 will have it |
| 10:10 | <annevk> | ta |
| 10:24 | <BenMillard> | krijnh, I converted the logs from a recent #css telecon as a demo for how "topic:" lines might create subheadings. Look at 6pm onwards, here: http://projectcerbera.com/!dev/irc-logs/day-telecon |
| 10:25 | <BenMillard> | krijhn, lines to/from you are coloured as if you were fantasai, to test how it looks when most lines get that colour. :) |
| 10:28 | <krijnh> | BenMillard: you're putting way too much time in this ;) |
| 10:29 | <zcorpan> | BenMillard: i almost see no difference between the light green and the white |
| 10:30 | <krijnh> | BenMillard: using "Topic: foo" as an indicator? |
| 10:32 | <BenMillard> | krijnh, yeah |
| 10:33 | <BenMillard> | zcorpan, some people can, some can't. I just copied the colour from the boxes in HTML5 spec. :) |
| 10:33 | <jgraham> | "OpenOffice.org requires a Java runtime environment" (JRE) to perform this task. Please install a JRE and restart OpenOffice.org" Usability FTW |
| 10:34 | <annevk> | http://www.fontsquirrel.com/ is nice |
| 10:34 | <jgraham> | (I have a JRE installed BTW) |
| 10:35 | <BenMillard> | zcorpan, I'm using a big CRT at 1600x1200 with low contrast and normal brightness, and I find it easy for me to see the difference. I can well imagine it's hard to tell the difference from laptops and maybe from TFTs. What colour looks better to you? |
| 10:36 | <BenMillard> | I've considered using the stringer green for "lines to/from you" and then use the :target orange for important lines...then find a new colour for :target lines |
| 10:36 | <BenMillard> | s/stringer/stronger/ |
| 10:36 | <zcorpan> | BenMillard: #dfd wfm |
| 10:36 | <annevk> | I suppose the main problem with most free fonts is that they do not cover the whole Unicode range |
| 10:36 | <annevk> | well, most expensive fonts don't either, but at least cover more |
| 10:37 | <zcorpan> | that would make them unusable as webfonts anyway |
| 10:37 | <annevk> | than just ASCII |
| 10:37 | <annevk> | zcorpan, not necessarily, you could subset them when you use them |
| 10:37 | <zcorpan> | (however for swedish purpuses åäö has to work) |
| 10:37 | <zcorpan> | annevk: ah ok |
| 10:37 | <zcorpan> | how? |
| 10:37 | jgraham | wonders how many fonts total cover the whole unicode range |
| 10:38 | <jgraham> | Like I guess the STIX fonts do if they ever actually released those |
| 10:38 | <annevk> | zcorpan, with some font editor; someone should probably write a web service when webfonts become more common |
| 10:38 | <annevk> | enter a bunch of characters and a font file and get a nice compressed version |
| 10:38 | <BenMillard> | zcorpan, that stays reasonably close to the spec. I'll try it in my user stylesheet for the current logs, maybe I'll prefer it. :) |
| 10:40 | <annevk> | BenMillard, what script do you use to detect code snippets? |
| 10:40 | <krijnh> | I don't think he does ;) |
| 10:41 | <BenMillard> | annevk, krijnh is correct. |
| 10:41 | <krijnh> | So the code snippets are a no-go for me |
| 10:41 | <annevk> | I see... |
| 10:41 | <BenMillard> | at the moment it's manual, based on searching for tell-tale sequences like space+:, word-word, <foo>, etc |
| 10:41 | <krijnh> | The rest is possible, I think |
| 10:41 | <krijnh> | But it takes a lot extra parsing for each line |
| 10:41 | <zcorpan> | krijnh: i'd like to see checkboxes instead of the current boxes |
| 10:42 | <BenMillard> | camelCase, foo:bar, foo(), foo.bar, things like this |
| 10:42 | <krijnh> | zcorpan: yeah, me too |
| 10:42 | <BenMillard> | the code snippets are a "blue sky thinking" kinda thing, I don't expect them to happen soon (or perhaps ever) |
| 10:43 | <krijnh> | Probably too buggy to automate |
| 10:44 | <annevk> | I think the hourly headings are nice |
| 10:44 | <BenMillard> | krijnh, too buggy to automate in a simple way, for sure. |
| 10:44 | <krijnh> | I think all the extra fluff is nice :) |
| 10:44 | <BenMillard> | annevk, thanks. Seemed useful for people who catch up on logs after knowing what time they left the previous day. |
| 10:45 | <BenMillard> | or for checking the proceedings of a meeting they knew started at a specific time |
| 10:47 | <annevk> | it seems the checkbox label should wrap around the entire line |
| 10:48 | <BenMillard> | annevk, I did that initially but it makes copy-paste of what people say have an unexpected side-effect. :) |
| 10:48 | <annevk> | true, then maybe have no label at all... |
| 10:49 | <BenMillard> | that makes the clickable area too small, imho. "Every engineering decision is a comprimise." |
| 10:50 | <BenMillard> | in Windows, selecting the name and the line does not tick the checkbox when only the name is a label |
| 10:50 | <BenMillard> | either the drag ends outside of the label, or it starts outside of the label |
| 10:50 | <annevk> | but for accessibility it seems better not to have labels than to have these |
| 10:50 | <annevk> | as hopefully heuristics for missing labels are better than what you give people now |
| 10:52 | <BenMillard> | annevk, I'm not sure how good they are now, but a year ago they were pretty basic. I could ask people to test it for me. |
| 10:52 | jgraham | wwishes that the ECMAScript people were not such fans of undefined behaviour |
| 10:58 | <BenMillard> | annevk, my current thinking is it's bettter to provide a short bit of labelling text because of Fitt's Law. This is also an accessibility advantage for users with mild mobility and dexterity problems. Screen reader users can move up and down the line if they want more context. I'll ask around. |
| 10:59 | <Philip`> | Couldn't you just make the checkbox larger, rather than adding a label, if you care about the clickable size? |
| 11:01 | <Philip`> | (Actually that's probably a stupid idea because it can't be any taller and it shouldn't be non-square) |
| 11:08 | <annevk> | http://twitter.com/ilinsky/status/1130145181 |
| 11:13 | <zcorpan> | jgraham: doesn't html5 actually allow an arbitrary amount of nested headers (just that the algorithm stops at the first just like gez says ATs do)? |
| 11:23 | <jgraham> | zcorpan: Does the table inspector behave differently? |
| 11:34 | hsivonen | reimplements the meta prescan by reusing the tokenization algorithm |
| 11:36 | <BenMillard> | ALA about mobile browsing recommends creating 4 stylesheets using '<link media>', '@import', '@import media' and media queries: http://www.alistapart.com/articles/returnofthemobilestylesheet |
| 11:36 | <annevk> | sounds like a lot of overhead :/ |
| 11:37 | <BenMillard> | annevk, apparently it's to work around web browsers ignoring media="handheld"...which makes me wonder how long it will be before they ignore media queries used this way, and then the next syntax for it, then the next, etc :) |
| 11:37 | <BenMillard> | web browsers on mobiles, i mean |
| 11:38 | <BenMillard> | I thought this was the behaviour mobiles would converge on: "2. Some read only the handheld style sheet if there is one, but default to the screen style sheet otherwise." |
| 11:39 | <annevk> | no, we're converging on just accepting media queries |
| 11:39 | <annevk> | and ignoring handheld altogether |
| 11:39 | <BenMillard> | oh, that's annoying |
| 11:39 | <annevk> | why? |
| 11:40 | <annevk> | way more likely people design properly for media queries, because they can easily test that on desktop |
| 11:40 | <annevk> | most handheld style sheets actually give a poor user experience |
| 11:41 | <zcorpan> | jgraham: no the inspector does what i expect |
| 11:41 | <BenMillard> | annevk, I've read that in several places but not seen examples of it. Are any documented? |
| 11:42 | <BenMillard> | annevk, "handheld" is easier for me to author. It is readable, easy to remember and leaves lets devices pick the stylesheet which suits their capabilities. |
| 11:42 | <BenMillard> | annevk, the media queries in that ALA article assume devices below a certain number of horizontal pixels can't do full-screen layout. |
| 11:43 | <annevk> | well, that seems buggy, mobile devices support the same CSS apart from some tat are flawed |
| 11:45 | <BenMillard> | annevk, in my experience mobile devices are highly diverse in what they support and how they present it. |
| 11:45 | <BenMillard> | annevk, better to let the device decide if it wants a linearised, simplified "handheld" stylesheet or if it can handle a desktop-like "screen" layout? |
| 11:45 | <zcorpan> | i tried to create a mobile layout for opera mini for my mom's site but i ended up putting it all in @media handheld because it just screwed up and the normal layout was more useful on mini anyway |
| 11:46 | <BenMillard> | I provide a linearising "handheld" stylesheet here, which works nicely on mobiles which have no zooming and apply the logic from "2" in the ALA article: http://calthorpepark.hants.sch.uk/ |
| 11:47 | <annevk> | BenMillard, the net result of that is that nobody wants handheld |
| 11:47 | <annevk> | BenMillard, users want a similar experience to desktop |
| 11:48 | <annevk> | I usually hate it when sites give me a limited mobile variant without way to opt out, e.g. klm.com does that and I need to URL hack them to get into the normal site |
| 11:49 | <BenMillard> | annevk, I hate that too, which is why I think the devices should decide based on <link media> rather than authors dictating it via media queries based on available width in pixels. |
| 11:50 | <BenMillard> | mobiles without zooming work better with linearised layouts. mobiles with zooming might have fewer available pixels across their width but still want the full layout |
| 11:51 | <zcorpan> | we might end up having to lie about the device pixels |
| 11:51 | <annevk> | we already do |
| 11:51 | <annevk> | about widths, etc. |
| 11:51 | <annevk> | of the viewport and such |
| 11:51 | <zcorpan> | wrt media queries? |
| 11:51 | <annevk> | i assume with media queries as well, because they should follow layout in principle |
| 11:51 | <Philip`> | People will probably design media=handheld stylesheets for the lowest common denominator of handheld devices, because they want it to work in as many as possible, but then anyone making a device slightly better than the lowest common denominator won't want to constrained by that stylesheet and would benefit from using the proper desktop one |
| 11:52 | <zcorpan> | i don't think we lie in media queries (based on my testing last week with mini) |
| 11:53 | <annevk> | k |
| 11:53 | <hsivonen> | how does Mobile Safari handle view port mqs? do they query the view port at readable zoom? |
| 11:53 | <hsivonen> | Mini does that, right? |
| 11:53 | <zcorpan> | i think so |
| 11:56 | <annevk> | http://www.paciellogroup.com/blog/?p=107 :/ how is that fricking better than HTML5? |
| 11:57 | <annevk> | oh, because hsivonen doesn't do landmark roles |
| 11:57 | annevk | sighs |
| 11:57 | <hsivonen> | annevk: "a problem with this is these rules do not allow ARIA landmark roles" |
| 11:58 | <hsivonen> | though if one starts hacking schemas, why hack a DTD? |
| 11:58 | <annevk> | because of validator.w3.org legacy? |
| 12:01 | <BenMillard> | "[...] I am not advocating the use of a custom DTD in published documents, but would like to be able to check my ARIA code." |
| 12:01 | <BenMillard> | sounds like he wants a QA tool, not a markup validator? |
| 12:01 | <jgraham> | BenMillard: A markup validator is a QA tool |
| 12:02 | <hsivonen> | BenMillard: there's demand for ARIA validators that'd silently accept anything as far as the non-ARIA bits go |
| 12:02 | <hsivonen> | which UAs support the landmarks and why don't they also support the corresponding HTML5 stuff? |
| 12:03 | <annevk> | to be honest, I don't like it that people already use <section> and <header> here and there |
| 12:03 | <annevk> | how are we going to adjust styling if adoption is there? |
| 12:03 | roc | discovers that xlink:href attributes in SVG <script> elements are animateable |
| 12:03 | <hsivonen> | annevk: I don't like it, either. |
| 12:04 | <annevk> | "SVG SUCKS" |
| 12:04 | <BenMillard> | jgraham, I meant a tool specially for checking and debugging ARIA, like hsivonen mentions there's demand for. |
| 12:04 | <BenMillard> | jgraham, something like a Javascript debugger more than a markup validator. |
| 12:04 | <annevk> | roc, though kidding aside, surely that's a bug? |
| 12:05 | <roc> | seems they deliberately made all href attributes animateable |
| 12:05 | <annevk> | sweet, with that and dynamic base URLs, headaches and crashes ensue |
| 12:06 | <BenMillard> | annevk, I guess the articles and presentations about HTML5 shouldn't be telling authors to try using the sectioning elements, then? |
| 12:06 | <roc> | there are fun scenarios about when script executes if a script file is loaded via animated href |
| 12:06 | <annevk> | BenMillard, either that or we should fix the styling problem and deploy it in browsers, preferably starting with something that has some user base, i.e. Gecko |
| 12:07 | <roc> | there are more fun scenarios about <use> with animated href causing the loading of external documents, possibly containing their own animations pointing back to the document containing the <use> |
| 12:08 | <hsivonen> | I still think <header> and <section> will mostly be failures without an outline depth-based selector |
| 12:08 | <zcorpan> | annevk: i think anything other than display:block shouldn't be in the UA's stylesheets for the new elements |
| 12:09 | <hsivonen> | zcorpan: the UA style sheet could adjust the size of hn nested in <section> based on the afore-mentioned depth selector |
| 12:09 | <zcorpan> | hsivonen: yeah maybe |
| 12:09 | <roc> | fortunately it appears that the href attribute of animation elements, identifying the element to be animated, cannot itself be animated |
| 12:10 | <zcorpan> | although maybe that would break author expectations (i use h6 because it's small and nice, now i switched to html5 markup it's as big as a h2!) |
| 12:10 | <hsivonen> | zcorpan: <section> would be the opt in to the new world |
| 12:12 | <zcorpan> | hsivonen: yes but my point is that maybe authors don't expect their headings to change size when they switch from div to section |
| 12:12 | <zcorpan> | if that's a good thing or not i don't know |
| 12:13 | <jgraham> | zcorpan: I think that would be good |
| 12:16 | <annevk> | maybe only for <h1> sizes should be adjusted |
| 12:17 | <jgraham> | annevk: That sounds bad because it means that there is a disconnect between the semantics and the style |
| 12:19 | <zcorpan> | it would have to be some disconnect anyway because we wouldn't want to change the size of <h6> outside <section> (which per html5 is equivalent to <h1>) |
| 12:19 | <hsivonen> | jgraham: separation of semantics and style! :-) |
| 12:20 | <zcorpan> | i.e. <body><h6>foo</h6><section><h2>bar</h2></section> |
| 12:23 | <jgraham> | Yeah, I guess the <section>-free case has to suck :( |
| 12:24 | <jgraham> | BenMillard: I would expect validtaor.nu to (eventually) validate the contents of aria-* |
| 12:24 | <jgraham> | Wow my typing sucks |
| 12:26 | <hsivonen> | jgraham: it already does, doesn't it? |
| 12:27 | <annevk> | there's <!doctype html>, <meta charset=utf-8>, <script> and <style> without type="", what else did we simplify? |
| 12:29 | <jgraham> | hsivonen: Er, I don't know. I assumed it didn't from what BenMillard said |
| 12:29 | <jgraham> | But I guess he was talking about something different |
| 12:30 | <BenMillard> | annevk, certain elements have a more lenient meaning and presentational elements are shown as being useful. That makes choosing a suitable element easier, imho. |
| 12:30 | <hsivonen> | annevk: profile |
| 12:30 | <jgraham> | Like a Firebug for aria |
| 12:30 | <hsivonen> | It would be great to have V.nu in Firebug. I just don't have the cycles to do it. |
| 12:31 | <BenMillard> | jgraham, a markup validator only tests the initial ARIA embedded in the markup. Checking the ARIA with scripting running enables proper troubleshooting and QA. This makes me think a specialised tool running in the browser (along the lines of Firebug and Dragonfly) are what ARIA authors would find most useful. |
| 12:31 | <BenMillard> | oh, what you both just said :P |
| 12:32 | <hsivonen> | porting Jing onto Google Web Toolkit should be doable, although compiling it would take an insanely long time |
| 12:32 | <BenMillard> | hsivonen, the HTML Validator Extension added the W3C validator as an extension to Firefox. Maybe that developer would be interested with integrating the HTML5 validation? http://users.skynet.be/mgueury/mozilla/ |
| 12:32 | <annevk> | hsivonen, profile is an open issue :) |
| 12:32 | <hsivonen> | perhaps it would require an switch in the GWT compliler to optimize *less* |
| 12:32 | <hsivonen> | BenMillard: perhaps |
| 12:33 | <zcorpan> | i wonder why steve hacked a dtd instead of hacking hsivonen's schema. perhaps validator.nu is a black box to him and he didn't want to approach hsivonen directly? |
| 12:33 | BenMillard | uses that extension all the time. |
| 12:34 | <hsivonen> | zcorpan: oh, I know it's an issue for steve, and I want to fix it--I just have so many things to do |
| 12:35 | jgraham | invokes cloneNode on hsivonen |
| 12:35 | <zcorpan> | hsivonen: ok. still, he could have hacked your schema instead of the dtd |
| 12:35 | <zcorpan> | jgraham: shallow clone? or with children? |
| 12:36 | <BenMillard> | does hsivonen have children? :) |
| 12:36 | <zcorpan> | dunno :) |
| 12:37 | <zcorpan> | applying the DOM api to humans is likely to be more messy than in IE |
| 12:41 | <jgraham> | zcorpan: Maybe that is why hsivonen has gone quiet |
| 12:42 | <jgraham> | I broke his internal state |
| 12:43 | <hsivonen> | :-) |
| 12:43 | <annevk> | jgraham, wasn't that bit undefined? :p |
| 12:44 | <annevk> | though then again, hsivonen is probably not a subclass of Node |
| 12:44 | <annevk> | to begin with |
| 12:55 | <zcorpan> | annevk: when will you implement the svgs on your blog? :) |
| 12:58 | <annevk> | likely tomorrow or tonight unless I finish my slides really quickly |
| 13:07 | <zcorpan> | cool |
| 13:10 | <Philip`> | "someone should probably write a web service when webfonts become more common" - has nobody already done that, even just for EOTs? It seems a fairly obvious thing for someone to provide |
| 13:13 | <annevk> | the EOT tool does that |
| 13:13 | <annevk> | but it's EOT specific |
| 13:13 | <annevk> | and not a web service |
| 13:14 | <annevk> | and EOT is not that popular |
| 14:45 | zcorpan | notices there's an HTML5+RDFa preset in validator.nu |
| 14:55 | jgraham | wasn't aware that it was easy to make the WHATWG and HTMLWG versions of the spec diverge |
| 14:55 | <jgraham> | That seems like a bad idea |
| 14:55 | <rubys> | Ian didn't say it was easy |
| 14:56 | <annevk> | I think it's quite easy |
| 14:57 | <annevk> | just annotate the section somehow and pre-process it |
| 14:57 | <annevk> | I'm not sure it's a good idea though |
| 14:57 | <jgraham> | Well I'm sure anolis can be made to do it. But it still seems bad |
| 14:58 | <annevk> | who objected to Origin exactly? |
| 14:58 | <annevk> | Origin is quite important and implemented in various clients already |
| 14:58 | <annevk> | and depends on HTML5 and vice versa |
| 14:58 | <rubys> | the suggestion is that it be pursued in the IETF |
| 14:59 | <jgraham> | By "easy" I meant "possible without modifying the spec generator code or maintaining two entirely different source files" |
| 14:59 | <annevk> | but HTML5 would still need to say that UAs include it for <form> submission |
| 14:59 | <rubys> | Perhaps Henri's response to the action is a bit incomplete? |
| 14:59 | <annevk> | jgraham, there is already some pre-process step before stuff goes to the spec generator |
| 15:00 | <annevk> | jgraham, removing a block seems trivial |
| 15:00 | <annevk> | rubys, I thought he summarized the situation quite accurately |
| 15:00 | <jgraham> | Oh, well maybe it is easier than I thought then |
| 15:00 | <rubys> | I didn't read in Henri's response that HTML5 depends on origin and vice versa. Going back to reread... |
| 15:00 | <jgraham> | Now there is just the question of whether it makes sense |
| 15:01 | <zcorpan> | why the rush to remove it anyway? |
| 15:02 | <rubys> | issue raised 2008-12-02. You call that a rush? |
| 15:03 | <annevk> | last publication, June 2008 |
| 15:07 | <zcorpan> | rubys: until it's in another spec i don't see why it should be removed |
| 15:07 | <annevk> | (which did not include this feature) |
| 15:15 | Philip` | wonders how much font files can be compressed, and discovers that Gentium contains 16KB of licensing information |
| 15:15 | <Philip`> | (It has the SIL Open Font License in ASCII, and then has it again in UTF-16) |
| 15:15 | <jcranmer> | no EBCIDIC? |
| 15:16 | <jgraham> | And it is now almost 14 years since the project started |
| 15:16 | <annevk> | sounds familiar |
| 15:22 | <annevk> | standards follow a similar schedule |
| 15:22 | <rubys> | not usually |
| 15:22 | <Philip`> | 14 years? Did they even have computers back then? |
| 15:22 | <rubys> | lol |
| 15:22 | rubys | mutters "kids" |
| 15:22 | <jgraham> | Philip`: I assume that's when they planned to carve the fonts into stone tablets |
| 15:22 | <Philip`> | OpenType is a bit complex :-( |
| 15:22 | <jgraham> | Although maybe that _is_ what they are doing given the time it is taking |
| 15:22 | <Lachy_> | rubys, I really don't like the idea of removing the Origin header from the HTMLWG copy while retaining it in the WHATWG copy. I agree with zcorpan's mail: there is no rush to remove it |
| 15:22 | <rubys> | Speak up on the list, I already said above that I'm willing to demote the status of the issue to "Raised". |
| 15:22 | <Philip`> | I have no idea who SIL are but I like how they develop decent free fonts and libraries |
| 15:22 | <rubys> | My issue isn't with the Origin header, my issue is with the cavalier approach the W3C takes with the issues list. |
| 15:22 | <Lachy_> | rubys, I would, but I don't have anything particularly new to add beyond what zcorpan said |
| 15:22 | rubys | looking for zcorpan's mail |
| 15:22 | <Lachy_> | http://www.w3.org/mid/op.un0fi6unidj3kv@hp-a0a83fcd39d2 |
| 15:23 | <rubys> | Hmmm. That didn't address the question. Thinking... |
| 15:25 | <rubys> | Thinking out loud: Spec'ed "somewhere" isn't the issue. The issue is scope of the html-wg. Things without consensus shouldn't appear in the W3C draft. I have zero control over what goes into the WHATWG draft. There appear to be no credible plans to ever republish the WHATWG document as a W3C draft at this time. Does that about cover the current status? |
| 15:27 | <Lachy> | rubys, Hixie has said multiple times that he isn't following the strictly concensus based approach for good reasons: it doesn't work; and I don't think we should start now |
| 15:29 | <rubys> | Noted. I strongly disagree, but noted. |
| 15:29 | <hallvors> | rubys, 'speced "somewhere"' is a big issue. It's called "avoid leaving things underspecified". |
| 15:29 | <Lachy> | I know you do, and I've noticed you've been trying to push concensus based approaches to resolve a few issues recently |
| 15:30 | <hallvors> | unfortunately using that as a principle bloats the HTML5 spec if noone else picks up loose ends :( |
| 15:30 | <rubys> | There are many, many, many areas in HTML5 over which there is strong agreement. I want those to proceed without being held up by areas over which agreement can not be reached. |
| 15:31 | <rubys> | Having a master document which contains everything and marks specific sections based on their status doesn't sound like such a bad idea given that as a context. |
| 15:32 | <Lachy> | rubys, the "can/can't live with" approach you've used for the Flash/Silverlight section and the DOCTYPE compat issue seems to be concensus driven based on nothing but personal preference and popular opinion. I really hope you can see what's wrong with that |
| 15:33 | <rubys> | I see the DOCTYPE compat issue as having a successful resolution. |
| 15:34 | <rubys> | I loved your email on the Flash/Silverlight topic. And believe that such an opinion would not have surfaced *unless* the issue was pushed as it has been. |
| 15:34 | <Lachy> | rubys, thankfully that discussion didn't get many "can/can't live with" responses, but was rather driven by evidence from Philip, newly proposed solutions and technical justifications |
| 15:35 | <rubys> | Thankfully? If anybody had tried to say "can't live with" without adequate justification, I would have pushed back strongly *until* there was evidence such as Philip provided. |
| 15:35 | <Lachy> | rubys, I don't have a problem with raising issues with the spec. I have a problem with the way you're attempting to resolve them by simply asking if people are ok with it, rather than actually encouraging people to look at the evidence |
| 15:36 | <rubys> | I am *not* simply asking them if they are OK with it. |
| 15:36 | <Lachy> | that's the impression I get from your emails |
| 15:36 | <rubys> | That leads to bikesheding. |
| 15:36 | <jgraham> | rubys: Maybe you are being too subtle |
| 15:36 | <Lachy> | perhaps you could express yourself more clearly |
| 15:37 | hallvors | googles bikeshedding |
| 15:37 | <jgraham> | To me "Can't live with" means something like "I would make a Formal Objection to the spec if this went in" |
| 15:38 | <rubys> | jgraham: agreed. And to me, a Formal Objection is a bit more than "I would prefer option 3 over option 2" |
| 15:38 | <jgraham> | So I could "live with" lots of bad things individually. But given too many of them the spec would be pretty crappy |
| 15:38 | Philip` | could live with HTML not existing at all |
| 15:38 | <rubys> | A perfectly wonderful outcome would have been five options that everybody could live with. I would simply have said in that case: editor, you are free to pick one. |
| 15:39 | <jgraham> | Pushing through to "Can't Live With" too early seems to have the strong possibility of preventng four of those five options from surfacing |
| 15:39 | <rubys> | But as it was, people not only said what they couldn't live with, but more importantly, they said why. And those reasons made sense. |
| 15:40 | <rubys> | jgraham: I don't like talking abstractly. I find too often that things that don't work in theory often work in practice, and vice versa. Did we get a successful outcome to the legacy compat issue? Was that resolution a result of me pushing? |
| 15:42 | <rubys> | And to be perfectly clear: I really would have been happy if nobody could have raised a "can't live with" objection to any of the listed alternatives. I don't know how I can say that any more clearly. |
| 15:42 | <jgraham> | rubys: So far we have no outcome to the issue :) But yes, in that case it looks like people pushing against the attitude of "Can't Live With" gave us a option that may be good enough to a problem with very little technical complexity |
| 15:42 | <rubys> | baby steps |
| 15:42 | <jgraham> | It's not clear what would happen in a different situation |
| 15:42 | <rubys> | let's find out |
| 15:43 | <Lachy> | the proposed resolution was a result of the issue being raised, but the resolution isn't necessarily final yet. There are still some issues with <!DOCTYPE html SYSTEM "about:sgml-compat"> that I would like to investigate |
| 15:43 | <Lachy> | in particular, how well do common consumers handle SYSTEM doctypes when they can't resolve the URI to a DTD |
| 15:43 | <rubys> | if you find an issue, please raise it. You can even raise an issue after Ian makes the change. |
| 15:44 | <Lachy> | I tested OpenSP briefly this morning and it seems to have some issues. I'm not sure if there's a way to either override the DOCTYPE or make it non-validating |
| 15:45 | zcorpan | tries validator.w3.org |
| 15:45 | <zcorpan> | "Fatal Error: cannot find "about:sgml-compat"; tried" |
| 15:45 | <rubys> | As I said in a prior email, Ian was willing to make the change last week. I asked that it remain open until at least Thursday for exactly this reason. |
| 15:45 | <hsivonen> | rubys: I don't know what deadline I should ask for. |
| 15:46 | <Lachy> | the validator would have to be updated to detect the new DOCTYPE in the same way it detects <!DOCTYPE html> |
| 15:46 | <rubys> | hsivonen: which issue? |
| 15:47 | <rubys> | I'm willing to demote issue 63 to raised |
| 15:47 | <zcorpan> | Line 1, Column 34: invalid formal public identifier "sgml-legacy": missing //. |
| 15:47 | <zcorpan> | <!doctype html public "sgml-legacy"> |
| 15:47 | <jgraham> | sgml//legacy? |
| 15:47 | <rubys> | the proposal was "about:sgml-legacy |
| 15:48 | <zcorpan> | rubys: yeah, i tried that first, got a fatal error |
| 15:49 | <Lachy> | rubys, actually, it was about:sgml-compat |
| 15:49 | <hsivonen> | rubys: the Origin header thing |
| 15:49 | <rubys> | sounds worthy of a "can't live with" objection. |
| 15:50 | <rubys> | hsivonen: that's issue 63. I'm willing to demote the issue to raised and leave the date open. |
| 15:50 | <hsivonen> | zcorpan: the goal is about compat with legacy producers. We aren't trying to be compatible with OpenSP consumers. |
| 15:51 | <hsivonen> | rubys: ok. one of the sensible approaches to the issue would entail leaving the references into the text anyway |
| 15:51 | rubys | notes that noting not all consumers can process this DOCTYPE will be a *heck* of a lot more effective than employing a pejorative text |
| 15:52 | <rubys> | such a doctype could even merit a validator warning |
| 15:56 | <Philip`> | Perhaps it's a problem that "can't live with" sounds like a personal view, and personally I don't care at all about XSLT or doctypes because they're not relevant to anything I'll ever do, but I'm still interested in finding technical issues that can inform the opinions of people who do care |
| 15:57 | <Philip`> | so rather than "can you live with this?", the question should be "can you live with this, or do you have any new information that might affect other people's views of whether they can live with this?", but that's not very concise really |
| 15:57 | <Lachy> | hsivonen, I was mostly interested in checking the DOCTYPE's compatibility with XML consumers, not text/html consumers, under the assumption that it could conceivably end up being used in XHTML documents despite it being unnecessary |
| 15:57 | <rubys> | I will quickly push back on people who attempt to say "can't live with" for purely spurious reasons |
| 15:58 | <Lachy> | rubys, can I suggest that instead of asking people to state whether they can or can't live with something, be more direct in asking for evidence for or against proposals |
| 15:59 | <rubys> | My experience is that that isn't quite as effective. It promotes bikeshedding. I simply want to exclude the options that don't work, and leave it up to the editor to select from those that will. |
| 16:00 | <annevk> | so far rubys approach seems to work fine |
| 16:00 | gsnedders | expects there will still be bikeshedding because the editor didn't choose x's preferred choice |
| 16:01 | <annevk> | I should probably say rubys' |
| 16:01 | <zcorpan> | rubys: a solution A might well work and might even be technically better than solution B even though someone can't live with A but everyone can live with B |
| 16:01 | <annevk> | zcorpan, but someone will only be heard if given good reason |
| 16:02 | <annevk> | afaict it's the same idea we've followed for ages with a new approach |
| 16:03 | <annevk> | and if that works better socially, lets go for it |
| 16:05 | <zcorpan> | if the decision in the end is exclusively based on the reasons then it's fine |
| 16:05 | <rubys> | if A, B, and C all work, I'm quite willing to let the editor do a coin flip |
| 16:06 | <Philip`> | I'm not sure that would work - the editor probably doesn't have a three-sided coin |
| 16:07 | <Lachy> | Philip`, he could roll a dice and assign 2 sides to each option |
| 16:07 | <annevk> | zcorpan, it just seems a bit conservative to object to rubys' approach at this point because so far it works |
| 16:08 | <jgraham> | Lachy: That wouldn't be a coin flip |
| 16:08 | <annevk> | or seems to work anyway, nothing has concluded yet |
| 16:08 | jgraham | suggests using 'heads' 'tails' and 'edge' |
| 16:08 | Philip` | also thinks you can't emulate a three-sided coin using multiple flips of a two-sided coin, because they don't have common factors, but he could be totally wrong |
| 16:08 | <zcorpan> | yeah, i'm not objecting to it, although the scenario i outlined might well occur |
| 16:08 | rubys | hates hypothetical scenarios |
| 16:09 | <hsivonen> | I imagine XSD has many things that people were able to live with |
| 16:09 | <jgraham> | Philip`: Flip if 6 times? |
| 16:09 | <annevk> | apart from the hypothetical note, how can someone not live with A without giving good reason? |
| 16:09 | <Lachy> | annevk, rubys' approach has obtained objections on the mailing list from both Hixie and hsivonen because it's wasn't clear that rubys really wanted justifications, rather than just running more like a vote |
| 16:09 | <rubys> | ... especially if it works well (as stipulated) |
| 16:09 | <annevk> | and if that reason is indeed good, why would everyone else like both? |
| 16:10 | <annevk> | Lachy, sure, new things always require some guidance |
| 16:10 | <zcorpan> | annevk: i didn't say he wouldn't provide a good reason |
| 16:10 | <zcorpan> | anyway |
| 16:10 | <annevk> | zcorpan, if he provides a good reason that would mean A might not be so good... I'd love some research and concrete examples here :p |
| 16:17 | rubys | notes that people are saying that they "can't live with" the "can't live approach" but haven't produced any evidence as to why it won't work |
| 16:18 | <rubys> | :-P |
| 16:18 | <Philip`> | jgraham: How would that work? You'd get 64 possible outcomes, and you can't evenly divide them into three classes |
| 16:18 | <zcorpan> | annevk: don't know any example off-hand but i'll let you know |
| 16:18 | <jgraham> | Philip`: Yes, I was being dumb |
| 16:18 | <jgraham> | Philip`: But can you do it if you don't distinguish options with the same set of outcomes, regardless of sequence? |
| 16:18 | karlcow | can live with everything as long if it's done in a good spirit, and with smooth social interactions. so far kudos to sam who did quite well. |
| 16:18 | <jgraham> | You maybe can't get all the probabilities to be equal :( |
| 16:18 | <Philip`> | jgraham: That would just be constructing groups out of the 64 events each with P=1/64, and you can't split them into three equal groups |
| 16:18 | <jgraham> | Yes, I realised that |
| 16:18 | <Philip`> | I suppose you could do it if you don't mind the potential for taking infinitely long |
| 16:18 | <jgraham> | It seems so odd that it isn't possible |
| 16:18 | <Philip`> | Flip 2 coins to get 4 outcomes, and if it's 1-3 then that's the answer, otherwise try again |
| 16:18 | <karlcow> | for what is worth, my coin has 3 sides |
| 16:18 | <danbri> | what's the preferred way for an html5 doc to declare charset? <meta charset="UTF-8"> ? |
| 16:18 | <annevk> | danbri, that and the old way are ok, HTTP is also fine |
| 16:18 | <jgraham> | Oh yeah. That would work. Now work out how many times Hixie will have to flip his coin on average and then determine the expected time for the two sides to wear away so they are indistinguishable |
| 16:18 | <karlcow> | danbri: <meta charset="UTF-8"> or <meta charset="UTF-8"/> or http headers |
| 16:18 | <annevk> | danbri, BOM as well, I don't think any of these is endorsed over the other |
| 16:18 | <annevk> | others, even |
| 16:18 | zcorpan | prefers http header |
| 16:18 | <jgraham> | karlcow: Not counting "edge"? |
| 16:18 | rubys | notes that meta charset is not currently endorsed for XHTML |
| 16:18 | jgraham | prefers <meta charset> |
| 16:18 | rubys | really wishes it were |
| 16:18 | <karlcow> | jgraham: I have two faces and one edge = 3 sides :) |
| 16:18 | <jgraham> | karlcow: Only really counts if you have a rather large edge :) |
| 16:19 | <karlcow> | euros, hk dollars for example work quite well, with two of them just now |
| 16:19 | <karlcow> | i18n ;) |
| 16:19 | <jgraham> | rubys: Having apparent encoding metadata that is actually not used in processing seems bad |
| 16:20 | <jgraham> | Alhtough I guess that's the model HTTP adopts |
| 16:20 | <Philip`> | jgraham: The average number of throws is just 1*(3/4) + 2*(1/4)*(3/4) + 3*(1/4)^2*(3/4) + ... I think, but I've forgotten how to calculate infinite series |
| 16:20 | <rubys> | it is used if intermediaries f*** with your headers. Don't say it doesn't happen. |
| 16:20 | <gsnedders> | Philip`: a/(1-r) |
| 16:20 | <gsnedders> | Philip`: (for a geometric sequence) |
| 16:20 | <Philip`> | gsnedders: It's not geometric |
| 16:20 | <gsnedders> | What sort of sequence? |
| 16:20 | <danbri> | i'll try <meta charset="UTF-8" /> for now |
| 16:20 | <gsnedders> | Philip`: Or what sequence? |
| 16:21 | <Philip`> | gsnedders: It's the one I said a few lines ago :-) |
| 16:21 | Lachy | wants to see karlcow's 3 sided coin (given the impossibility of creating a 3 sided, 3 dimentional shape) |
| 16:21 | <jgraham> | sigma(n(3/4)^n) |
| 16:21 | <karlcow> | lachy? |
| 16:21 | <Dashiva> | Lachy: It's just a cylinder |
| 16:21 | <karlcow> | a coin has 3 sides |
| 16:21 | <karlcow> | a cylinder |
| 16:21 | <Philip`> | Fortunately Perl can calculate near-enough-to-infinite series and tells me the answer is 4/3 |
| 16:22 | <Philip`> | karlcow: A coin only has one side (the outside) |
| 16:22 | <karlcow> | or if you indulge my French, a camembert |
| 16:22 | <Lachy> | Dashiva, ah, I meant flat sides |
| 16:22 | <Lachy> | you can't land on the edge of a coin in practice |
| 16:23 | <Dashiva> | If it's a really fat coin :) |
| 16:23 | <Philip`> | You could glue many coins together, until it lands on the side roughly a third of the time |
| 16:23 | <karlcow> | :) |
| 16:23 | <rubys> | you can solve this easily yourself: write down "x = sigma(n(3/4)^n)" Next write "(3/4)x = sigma(n(3/4)^(n+1))", and subtract the two equations (it is helpful to expand the sigma doing this". Then simply solve for x. |
| 16:23 | <Philip`> | but it'd probably be really easy to cheat by throwing it so it spins along its axis and always lands on its edge |
| 16:24 | <karlcow> | Half a cork would do |
| 16:24 | Philip` | notes that it's actually sigma(n * 1/4^(n-1) * 3/4) |
| 16:24 | <Philip`> | (not that it makes much difference in practice) |
| 16:25 | <karlcow> | plus the probability will become more and more interesting each time you have to open a bottle :) |
| 16:26 | <Philip`> | rubys: Oh, that seems to work quite easily |
| 16:27 | <Philip`> | (since it gives 3/4x = a geometric series which sums to 1) |
| 16:27 | <rubys> | if you select the right term, you end up with fraction*x = a simple value |
| 16:28 | <rubys> | what you select is the term which will cause each entry in the series to shift over by one. |
| 16:28 | <Philip`> | I don't see how that's possible in this case |
| 16:29 | <Philip`> | (where 'this case' is basically x = r + 2r^2 + 3r^3 + ...) |
| 16:29 | <karlcow> | hmmm not a new question (as usual) https://home.comcast.net/~davejanelle/coin3.htm |
| 16:30 | <Philip`> | (so if you shift it and subtract then you're still left with an infinite series but you lose all the 2* and 3* bits) |
| 16:30 | <rubys> | how did that n get in there? you have 3/4 of a chance at first. Then you have 3/4 of a chance for the remaining 1/4 of possibilities. |
| 16:31 | <Philip`> | The aim is to calculate the average number of throws needed |
| 16:31 | <karlcow> | http://www.statisticool.com/3sided.htm |
| 16:31 | <Philip`> | so there's a 3/4 chance of needing 1 throw, 3/4*1/4 of needing 2 throws, etc |
| 16:31 | <rubys> | ah, nevermind |
| 16:32 | <Philip`> | Otherwise it'd just be calculating the probability that you eventually stop, which is fortunately approaching 1 :-) |
| 16:32 | <rubys> | ok, so the technique needs to be applied twice, but you shortcutted it be simply recognizing the sequence. |
| 16:32 | <karlcow> | http://www.cs.umass.edu/~scottk/files/threesidedcoin_amstat.pdf |
| 16:33 | <karlcow> | article Teaching Bayesian Model Comparison With the Three-Sided Coin by Scott Kuindersma and Brian Blais in The American Statistician, August 2007, Vol. 61, No. 3. |
| 16:33 | <karlcow> | unfortunately dead link |
| 16:34 | <Philip`> | rubys: I don't remember having seen that trick for solving series - I think I was just told how to solve arithmetic and geometric series and not much else |
| 16:34 | <Philip`> | karlcow: http://www-stat.stanford.edu/~owen/courses/306a/threesidedcoin_amstat.pdf is less dead |
| 16:34 | <karlcow> | http://bryant1.bryant.edu/~bblais/pdf/pres_011407.pdf working link here too |
| 16:35 | <karlcow> | thanks Philip` |
| 16:35 | <rubys> | I was a math major, and have now helped two teenagers though this (the oldest no longer being a teenager) |
| 16:36 | Philip` | is a computer scientist so he doesn't have to know any maths that has actual numbers in it |
| 16:36 | <Philip`> | (and especially not anything with non-integers) |
| 16:39 | <rubys> | I was a math major because when I went to school, there wasn't a computer science degree I could have pursued. |
| 16:41 | <karlcow> | nice article |
| 16:43 | <Philip`> | rubys: If you want to do HTML WG mail from Gmail, before the W3C gets around to changing your registered email address, you could possibly just forward it to Gmail and reply directly from there, since the mailing list accepts mail from any address (not just registered members) |
| 16:44 | <Philip`> | (I sort of do the opposite - my Gmail address is registered on the list, but I can read it through IMAP in Thunderbird and then I reply from a different address instead) |
| 16:47 | <jcranmer> | Philip`: no IEEE 754 floating point? |
| 16:49 | <Philip`> | jcranmer: Not as part of the CS stuff I'm working on |
| 16:50 | <Philip`> | (I think in the past year, the extent of discussion around floating point was that someone commented that FP addition was non-associative and we don't know how to handle that in our theoretical model of the stuff we're working on so we'll ignore it for now) |
| 16:51 | <Philip`> | And floating point numbers are still really just integers anyway :-p |
| 16:52 | <jcranmer> | simple truth of computers: mathematical truths don't necessarily hold true when computers are involved |
| 16:54 | <rubys> | Since their email form doesn't work, and my [sysreq] submitted on Friday hasn |
| 16:54 | <rubys> | 't worked, I just submitted a request for a second id. |
| 16:54 | <rubys> | we'll see how that goes. |
| 16:55 | <Philip`> | It's just a bit confusing when a+(b+c) != (a+b)+c |
| 16:56 | <Philip`> | (or at least it's a bit confusing to try to prove things about systems that involve algebras with that property) |
| 16:58 | <jgraham> | hallvors: I think you are much more optimisitic about the utility of enhanced copy/paste than me. |
| 16:59 | <hallvors> | I think the web isn't very good at being copied. |
| 16:59 | <hallvors> | :-p |
| 17:00 | <hsivonen> | hallvors: didn't MS have something like that as "active clipboard" or similar Microsofty word plus "clipboard"? |
| 17:01 | <hallvors> | the "copy E-mail from webmail to mail client" example was somewhat far off, admittedly. but the bibliography thing could be done today and would be GREAT for anyone using web for research and word processors for writing essays and dissertations. |
| 17:01 | <hsivonen> | http://news.cnet.com/Microsofts-CTO-envisions-a-Web-savvy-clipboard/2100-7345_3-6047029.html |
| 17:01 | <hallvors> | live clipboard |
| 17:01 | <rubys> | http://www.oreillynet.com/conferences/blog/2006/03/ray_ozzie_wires_the_web_with_a.html |
| 17:01 | <hallvors> | I looked for that but the demos and such were not available now |
| 17:01 | hsivonen | forgot that MS things are now live rather than active |
| 17:01 | <jgraham> | On the one hand I can imagine producers not wanting to give up their data e.g. Facebook not wanting to make it easy to migrate data away from facebook |
| 17:01 | <rubys> | it was originally "universal" |
| 17:02 | <jgraham> | On the other hand I think complex things happening on "paste" break user expectations e.g. pasting from wikipedia shouldn't do anything other than insert the text |
| 17:02 | <Philip`> | People who use Facebook are too cool to use email, and people who copy-and-paste from Wikipedia into a word processor don't want their teacher to know they copied it |
| 17:03 | <hallvors> | jgraham: I sync my Facebook contact list with my local address book *manually*.. it would be a competitive advantage to any social network to make that easier for me. |
| 17:03 | <hallvors> | (if only they'd realise :-) ) |
| 17:04 | <jgraham> | hallvors: But it is a competitive advantage to them not to as long as they are the market leader |
| 17:04 | <hallvors> | pasting into word processors these days usually gives you a floating palette toolbar thingy with options. I don't really like those but if one of the options were "create bibliographic entry for what you pasted" I'd use it a lot |
| 17:04 | <hallvors> | ..and we're not writing HTML for the kiddies who cheat at school.. |
| 17:05 | <hallvors> | not if e.g. myspace or orkut does it |
| 17:06 | <hallvors> | then it might suddenly be a competitive advantage to have that feature |
| 17:06 | <hallvors> | I need to sleep anyway. |
| 17:08 | <karlcow> | hmmm |
| 17:08 | <karlcow> | hallvors: facebook to your computer addressbook or the opposite? |
| 17:08 | <hallvors> | metadata stuff gets us quite deep into "UI problem" territory... |
| 17:08 | <hallvors> | karlcow: why not both? |
| 17:09 | <karlcow> | ethics |
| 17:09 | <karlcow> | the data about people in my addressbook doesn't belong to me. People trusted me giving their information. |
| 17:09 | <hallvors> | Facebook would detect an onpaste event and try to figure out the clipboard data |
| 17:10 | <karlcow> | and I don't feel the right to share these data with a 3rd party with a contract which makes the possibility to use these data |
| 17:10 | <hallvors> | if somebody gave you their E-mail address and you use it to search for them on facebook is it ethically wrong?? |
| 17:11 | <karlcow> | if I don't know the retention policy of the service about this feature. yes it's wrong |
| 17:11 | <karlcow> | in my own way of dealing with data |
| 17:11 | <karlcow> | :) |
| 17:12 | <hallvors> | so will you take this to the logical conclusion that we should not enable any sort of data exchange because people might use it to expose data others gave them to thirdparties? :-o |
| 17:13 | <hallvors> | You need to find the Internet's off switch.. |
| 17:13 | <karlcow> | hallvors: extreme position never solves things |
| 17:13 | <karlcow> | my simple take on it is "it depends on the context" |
| 17:13 | <karlcow> | for this we need mechanisms for knowing how we can use the data. |
| 17:14 | <hallvors> | I think it's a bit extreme to say it's wrong of me to search for you on Facebook by your E-mail.. |
| 17:14 | <karlcow> | email, addresses and phone numbers of my friends are not *my* data. There are data which have been communicated under a moral contract (trust) |
| 17:14 | <karlcow> | hallvors: did I say that |
| 17:15 | <karlcow> | hallvors: but yes if you put my email address or phone number on facebook I would consider that wrong. |
| 17:15 | <karlcow> | plus I'm not on facebook ;) |
| 17:15 | <hallvors> | with a caveat about retention policy, you said "yes, it's wrong" |
| 17:15 | <Philip`> | Technical solutions to moral/social problems are probably not going to be particularly effective |
| 17:15 | <karlcow> | Philip`: you jumped to your own conclusions ;) not what I said |
| 17:17 | <karlcow> | ex of technology: creative commons licenses markup. The fact after that you respect the license or not is another issue completely |
| 17:17 | <karlcow> | but at least there is a license |
| 17:18 | <Philip`> | There isn't technology that attempts to enforce that licence by preventing you from copying-and-pasting into a document with an incompatible licence |
| 17:18 | <karlcow> | the thing is that often right now there is no way for some data owners in some context to declare by technology "you can't use the data for this" |
| 17:18 | <Philip`> | (which sounds like what you're suggesting for preventing copying-and-pasting from your address book into Facebook) |
| 17:18 | <karlcow> | :) then sure you can decide to break or not |
| 17:19 | <karlcow> | not preventing, declaring |
| 17:20 | <karlcow> | when I give my phone number to a friend (verbally), I usually give a comment on the type: do not share this number with someone else or ask me before if you can do it. |
| 17:20 | <karlcow> | same thing for my home address |
| 17:20 | <hallvors> | If I search for your address on Google Maps I'm giving some of your personal data to Google. yet very few people would consider that an immoral thing to do.. |
| 17:20 | <rubys> | few people realize how much google knows about them |
| 17:20 | <rubys> | http://www.google.com/search?q=sam+ruby+nc |
| 17:21 | <karlcow> | yes sam, only a few people, and not much effort is done in UI to make them aware of it. (killing business I guess) |
| 17:23 | hsivonen | wonder if Google has enough data to make the voter registration thing obsolete |
| 17:23 | <hsivonen> | *wonders |
| 17:24 | hallvors | is amused by Sam's post about lying to his GPS to get a correct route home |
| 17:25 | <hallvors> | Actually, these days Facebook doesn't need to support whatever-meta-data-embedding for such a proposal to be useful. |
| 17:25 | <rubys> | looks like google maps at least has fresh data so lying eventually won't be necessary |
| 17:25 | <hallvors> | If the UA does some clever copy thing, one can have a user script / greasemonkey script that knows the target site's markup and creates the meta data |
| 17:25 | <hsivonen> | hmm. Microsoft has fixed the bug I posted in Sam's comments |
| 17:26 | <hallvors> | in Japan GPS navigation seems very good. |
| 17:26 | <rubys> | which bug? |
| 17:27 | <hallvors> | ..but then, they REALLY need it around here.. |
| 17:27 | <hsivonen> | rubys: the link at http://intertwingly.net/blog/2005/04/06/Dude-Wheres-My-Car#c1112998015 used to lead to a route search from one part of a Finnish to another part of the same city and MSN returned a very crazy route (IIRC traveling through Sweden or something like that) |
| 17:27 | <karlcow> | hallvors: not GPS, more likely celltowers triangulation because of the density of mobile phones. GPS doesn't work very well in the cities ;) no more sky |
| 17:28 | <hsivonen> | IIRC, it suggested driving to northern Sweden, driving through Sweden, crossing over to southern Finland and driving back north |
| 18:48 | gsnedders | doesn't think he's ever done so badly in a physics exam |
| 19:13 | <Philip`> | gsnedders: I thought the same in a maths exam once, and then scored 100% on it (after normalisation of scores), so there's not necessarily much correlation between perception and reality :-) |
| 19:14 | <Dashiva> | It was fun getting more than 100% on tests due to extra credit (when I lived in Seattle) |
| 19:14 | <heycam> | roc, that was corrected in 1.2T (@xlink:href on <script> being animatable) |
| 19:15 | <Dashiva> | One time I got extra credit on an extra credit question for spotting a mistake |
| 19:16 | <gsnedders> | Philip`: I got 54%. I thought I did well. :) |
| 19:17 | <Philip`> | gsnedders: Oh, if you actually got told the score then I suppose that's correlated with reality |
| 19:17 | <gsnedders> | Philip`: I thought I had probably got a low A, not a low C. |
| 19:18 | <gsnedders> | I also got a B in computing, by not knowing what the right, sorry, wrong answers are. |
| 19:18 | <Philip`> | That sounds unfortunate |
| 20:18 | Philip` | tries to hack together a font subsetter |
| 20:19 | <Philip`> | When used for a dozen ASCII letters, a font like Gentium goes from 260KB to 25KB, of which 13KB is the licence text |
| 20:21 | <Philip`> | and annevk's big-noodle-titling goes from 26KB to 8KB, of which 4KB is the licence text |
| 20:22 | <Philip`> | (I've probably got lots of bugs that break kerning and everything, though) |
| 20:27 | <hsivonen> | Philip`: all browsers that support @font-face support gzip, don't they? |
| 20:27 | <hsivonen> | Philip`: wouldn't it be safe to serve a gzipped font unconditionally without conneg? |
| 20:27 | <hsivonen> | presumably that would take care of the license text |
| 20:28 | <Philip`> | With gzip, it's 3.5KB for big-noodle-titling and 11KB for Gentium |
| 20:30 | <hsivonen> | I want to try how hard it will be to use Linux Libertine, MgOpen Cosmetica and Libertine Mono with @font-face with strict license compliance |
| 20:31 | <hsivonen> | i.e. providing full corresponding source code |
| 20:32 | <Philip`> | Removing the copyright text and using gzip, it's 2.7KB and 7KB respectively |
| 20:33 | <hsivonen> | Philip`: does Gentium allow the removal of the license text and replacing it with the URL of the license? |
| 20:33 | <takkaria> | what constitutes source code in the case of fonts? |
| 20:33 | <Philip`> | http://scripts.sil.org/cms/scripts/render_download.php?site_id=nrsi&format=file&media_id=OFL_plaintext&filename=OFL.txt |
| 20:34 | <hsivonen> | takkaria: the preferred form for making modification, which is the FontForge input file |
| 20:34 | <Philip`> | "Original or Modified Versions of the Font Software may be bundled, redistributed and/or sold with any software, provided that each copy contains the above copyright notice and this license. These can be included either as stand-alone text files, human-readable headers or in the appropriate machine-readable metadata fields within text or binary files as long as those fields can be easily viewed by the user." |
| 20:35 | <hsivonen> | someone should make a font serving site that crawls the customer sites and subsets fonts automatically |
| 20:37 | <zcorpan> | hsivonen: do you mean that it would subset it to include characters found in the text that uses the font? |
| 20:37 | <hsivonen> | zcorpan: yes |
| 20:37 | <hsivonen> | so the crawler would need to be CSS-aware |
| 20:38 | <takkaria> | hmm, that wouldn't be too hard |
| 20:38 | <hsivonen> | takkaria: except with JS adding more text :-) |
| 20:38 | <zcorpan> | or a comment form |
| 20:39 | <takkaria> | JS makes the idea unusable, I guess |
| 20:39 | <Philip`> | I was idealistically unrealistically thinking of making something vaguely like that |
| 20:40 | zcorpan | would be happy if he could choose which characters to include from a character map or pasting the characters he wanted |
| 20:41 | <zcorpan> | e.g. if i wanted to use a webfont for my mother's site's headings i'd want swedish and icelandic characters |
| 20:45 | Philip` | presumes the main concern for someone providing such a service would be the licensing of fonts |
| 20:46 | <hsivonen> | Philip`: one would only offer to serve OFL and GPL fonts |
| 20:46 | <hsivonen> | and Apache License 2.0 fonts |
| 20:46 | <Philip`> | And not let people upload their own fonts? |
| 20:46 | <hsivonen> | right |
| 20:46 | <roc> | TTF fonts have an "installable" bit |
| 20:47 | <roc> | which basically means "free" |
| 20:47 | <roc> | a site could allow file uploads if it checked that bit |
| 20:47 | <Philip`> | Does that imply it allows derivative works too? |
| 20:47 | <zcorpan> | YouFont |
| 20:48 | <zcorpan> | unsurprisingly there's such a site |
| 20:48 | <zcorpan> | (named youfont i.e.) |
| 20:48 | <zcorpan> | (although it doesn't seem to be the same concept as youtube - i don't see any upload) |
| 20:56 | <Lachy> | hsivonen, if making the source of the font available is a requirement, then surely a URL for it in a CSS comment near the @font-face rule, or some other appropriate location, would be acceptable |
| 20:58 | <Lachy> | it the same with copyleft licenced software. You don't need to distribute the source with the binary, just make it available to those who want it |
| 21:09 | <hsivonen> | is this for real: http://twitter.com/zacharyjohnson/statuses/1131296427 |
| 21:14 | <roc> | I wonder if http://alistapart.com/articles/semanticsinhtml5 is for real |
| 21:14 | <roc> | it sounds like another form of <tag name="section"> |
| 21:18 | <Lachy> | it doesn't surprise me that there are still people out there hoping for the day that HTML will be replaced with semantically pure, XML based solutions |
| 21:22 | <hsivonen> | Lachy: it's not the XML part. it's the DTD part. |
| 21:23 | <hsivonen> | XML standards types have moved to RELAX NG, Schematron and NVDL |
| 21:23 | <hsivonen> | some are sticking to XSD, but DTD is really passé |
| 21:28 | <zcorpan> | http://www.webfonts.info/wiki/index.php?title=Fonts_available_for_%40font-face_embedding |
| 21:31 | <Philip`> | Is Bitstream Vera / DejaVu compatible with embedding? |
| 21:31 | Philip` | didn't see anything in its licence that indicated otherwise |
| 21:31 | <hsivonen> | Philip`: seems linkable and embeddable to me |
| 21:32 | <hsivonen> | the list on webfonts.info is very incomplete as far as OFL and GPL fonts go |
| 22:14 | zcorpan | notes that annevk's big-noodle-titling is served as text/plain |
| 22:16 | <zcorpan> | (and so is the webfont i uploaded for daddy.svg) |
| 22:29 | <Lachy> | what's the correct MIME type for a font? |
| 22:30 | <zcorpan> | i think there is no yet, but it's being discussed... i used font/ttf |
| 22:31 | <Lachy> | ah, yeah, I remember http://annevankesteren.nl/2008/08/font-mime-types |
| 22:34 | <Philip`> | How do I make Opera stop using my system font instead of the @font-faced version that I want it to use? |
| 22:35 | <Philip`> | (It works more sensibly in Firefox, where it falls back to the page's default font for missing glyphs) |
| 22:36 | <Lachy> | Philip`, can you demonstrate the problem? |
| 22:37 | <Philip`> | Lachy: Not easily |
| 22:43 | <Philip`> | How does Opera pick fallback fonts? |
| 22:43 | <Philip`> | I changed all the name strings in the font, so it can't just be matching those |
| 22:51 | <zcorpan> | annevk: hey |
| 22:51 | <Lachy> | Philip`, can you publish a test case of some kind? |
| 22:51 | <annevk> | http://twitter.com/Kroc/statuses/1131482645 |
| 22:51 | <annevk> | zcorpan, yo |
| 22:51 | <annevk> | zcorpan, not going to happen tonight I'm afraid |
| 22:51 | <annevk> | zcorpan, the SVG stuff |
| 22:52 | <Lachy> | annevk, yeah, that twitter message is a good reason to retain <ol type> |
| 22:52 | <zcorpan> | annevk: i just did it myself -- wanted to see how it would look :) |
| 22:52 | <annevk> | I see :) |
| 22:52 | <zcorpan> | although it looks terrible in safari and firefox (because they don't support svg in content) |
| 22:53 | <zcorpan> | should i upload it? |
| 22:53 | <zcorpan> | i also got rid of fake-border |
| 22:56 | <takkaria> | more feed readers need css sanitisers |
| 22:57 | <Philip`> | Lachy: http://philip.html5.org/tests/font/gentium-embedded.html |
| 22:58 | <Philip`> | Lachy: probably with http://www.sil.org/~gaultney/Gentium/ already installed on your system |
| 22:58 | <Philip`> | http://scripts.sil.org/cms/scripts/render_download.php?site_id=nrsi&format=file&media_id=GentiumBasic_110&filename=GentiumBasic_110.zip in particular |
| 22:59 | <Philip`> | Lachy: In Opera 10 build 4125 on Linux, the whole text is rendered in Gentium |
| 22:59 | <roc> | zcorpan: we don't support what what? |
| 22:59 | <Philip`> | Lachy: whereas in Firefox 3.1 only the e/m/b/d letters are Gentium and the rest is default sans-serif, which is what I want it to be |
| 22:59 | <Philip`> | s/want/expect/ |
| 23:00 | <zcorpan> | roc: css 'content' |
| 23:00 | <Lachy> | works for me |
| 23:01 | <roc> | zcorpan: er |
| 23:01 | <Lachy> | using Opera 10 Mac, build 6196 |
| 23:01 | <zcorpan> | roc: e.g. :before { content:url(foo.svg) } |
| 23:01 | <roc> | oh, SVG images |
| 23:01 | <roc> | OK |
| 23:01 | <annevk> | zcorpan, sure |
| 23:01 | <Lachy> | it may be a bug an the older build that's now fixed |
| 23:01 | <hsivonen> | Philip`: WFM in Opera 10 on Mac |
| 23:01 | <roc> | yeah, that's next on my list |
| 23:01 | <roc> | but that should work in Webkit |
| 23:01 | <Lachy> | hsivonen, which build have you got? |
| 23:01 | <zcorpan> | annevk: http://simon.html5.org/dump/anne.html |
| 23:02 | <hsivonen> | Lachy: 6166 |
| 23:02 | <annevk> | oh, Firefox does do the hand? |
| 23:02 | Philip` | tries twiddling the test a bit |
| 23:02 | <annevk> | oh wait, that's Opera |
| 23:03 | <zcorpan> | annevk: i changed the search form stuff a bit to make it look right |
| 23:03 | <zcorpan> | maybe i broke it in some other browser |
| 23:03 | <Lachy> | it could be a linux specific bug. Who else has linux here that can test the latest internal builds? |
| 23:03 | <Philip`> | Oh, how odd |
| 23:03 | <Philip`> | I don't actually have Gentium installed, as far as I can tell |
| 23:03 | <Philip`> | so my hypothesis was faulty |
| 23:03 | <annevk> | zcorpan, the logo is not updated? |
| 23:04 | <annevk> | uploaded |
| 23:04 | <Philip`> | so the question is: where is Opera getting the glyphs for all the other characters from? :-/ |
| 23:04 | <zcorpan> | annevk: i haven't used sam's version of the logo |
| 23:06 | <zcorpan> | Philip`: maybe you have used a different webfont with the same name that opera cached? |
| 23:06 | <Philip`> | Oh |
| 23:06 | <Philip`> | I'm being stupid |
| 23:06 | <Philip`> | It's not actually using Gentium, it's just using something that looks very similar |
| 23:07 | <Lachy> | I have no idea what Gentium is. Is that the name of the font you're embedding? |
| 23:07 | <Philip`> | (Somehow it's picking "Bitstream Charter" as the fallback, and I have no idea why) |
| 23:07 | <Philip`> | Lachy: Yes - see the links I posted some minutes ago |
| 23:07 | <Lachy> | is Bitstream Charter a serif font? |
| 23:08 | <Philip`> | Yes |
| 23:08 | <Lachy> | ok. That is weird. It should be picking a sans-serif fallback font |
| 23:08 | <hsivonen> | Philip`: in case you are wondering, I do have Gentium installed |
| 23:09 | <Philip`> | http://philip.html5.org/tests/font/gentium-embedded-opera.png |
| 23:10 | <zcorpan> | annevk: the logo is uploaded... doesn't it work? |
| 23:10 | <Philip`> | It's more obvious now than when I initially had 1em text :-( |
| 23:10 | <Lachy> | Philip`, if you remove the @font-face, what fallback font does it use? |
| 23:10 | <Philip`> | Lachy: A sans-serif one |
| 23:11 | <annevk> | now it does |
| 23:12 | <annevk> | still usable in Firefox |
| 23:12 | <annevk> | guess I'll use it then just for fun |
| 23:13 | <zcorpan> | :) |
| 23:13 | <annevk> | though hopefully rubys can make the files somewhat smaller or maybe I should learn to use his tools + SVG |
| 23:13 | <zcorpan> | he uploaded a smaller version of the logo at least |
| 23:13 | <Philip`> | Hmm, fonts are fun - I have one which looks terribly ugly in Opera, but the spacing in the subsetted version is correct, whereas it looks pretty in Firefox but the subsetted version has all its characters overlapping |
| 23:14 | <zcorpan> | annevk: http://intertwingly.net/tmp/logo3.svg |
| 23:15 | <annevk> | k |
| 23:15 | <annevk> | will still do it "tomorrow" though |
| 23:15 | annevk | is tired |
| 23:16 | <zcorpan> | that's ok |
| 23:16 | <annevk> | good :) |
| 23:16 | <zcorpan> | :) |
| 23:20 | <Philip`> | Aha, Firefox is happy when I fix the hdmx table |
| 23:20 | <Philip`> | which Opera presumably ignores |
| 23:21 | <zcorpan> | annevk: also, you might want to add AddType font/ttf .ttf |
| 23:23 | <annevk> | zcorpan, is anyone fixing the media type issue then? |
| 23:23 | <zcorpan> | annevk: dunno |
| 23:24 | annevk | tried, didn't get enough support and got distracted |
| 23:25 | <zcorpan> | currently it's text/plain |
| 23:25 | <annevk> | I mean fixing the media type issue in browsers |
| 23:26 | <annevk> | I know it's broken on my server, I didn't specify anything at all |
| 23:27 | <annevk> | I wonder why Larry moved the Origin thread from WebApps to the HTML WG :/ |
| 23:27 | <annevk> | I also wonder why he claims that the W3C cannot introduce new headers and protocols, they can |
| 23:31 | <annevk> | lol http://my.opera.com/desktopteam/blog/reminder-opera |
| 23:36 | <Philip`> | Is there a way to disable spellchecking in <p contenteditable> in Firefox and Opera? |
| 23:36 | <Philip`> | (The red underlines are ugly) |
| 23:36 | <annevk> | spellcheck="no" in Firefox? |
| 23:37 | <Philip`> | Doesn't seem to work |
| 23:37 | <annevk> | oh, I thought they had that implemented |
| 23:37 | <Philip`> | Maybe only on form fields? |
| 23:38 | <annevk> | maybe :) |
| 23:38 | <zcorpan> | Philip`: iirc, the spellcheck implementation predated contenteditable in firefox |
| 23:39 | Philip` | is using the latest FF3.1 nightly |
| 23:44 | <roc_> | spellcheck should work with contenteditable I thought |
| 23:46 | <Philip`> | I must be doing something stupid because I still get wavey red lines |
| 23:48 | <annevk> | maybe it's spellcheck="false" ? |
| 23:48 | annevk | isn't sure about the exact syntax |
| 23:48 | <Philip`> | I tried that too |
| 23:48 | <annevk> | ah, it's "off" |
| 23:49 | <annevk> | just like autocomplete I suppose |
| 23:50 | <annevk> | did Microsoft invent both autocomplete and contenteditable? |
| 23:50 | <roc_> | I think we must have broken it somehow |
| 23:51 | <Philip`> | data:text/html,<p contenteditable spellcheck="off">Slartibartfast |
| 23:51 | <Philip`> | underlines in red in FF |
| 23:51 | <Philip`> | and also in Opera 10 once I click on the text |
| 23:51 | <roc_> | yeah, something's busted |
