| 00:02 | <Philip`> | mpt: Oh, thanks - that looks like it would be interesting if I had Java installed :-) |
| 00:20 | <tndH_> | http://blogoscoped.com/archive/2006-07-28-n51.html is a lot less thorough, but it's colourful :) |
| 00:23 | <mpt> | I wonder how much the frequencies of "9" and "11" have increased in the past decade |
| 03:25 | <Hixie> | why i do believe Leif just asked me to shut up |
| 03:33 | <karlUshi> | paranoia. |
| 03:33 | <karlUshi> | I think Leif said that 80%-20% was a poor argument in a discussion. |
| 03:33 | <Hixie> | "Save it up for later." seems pretty clear to me |
| 03:34 | <karlUshi> | yes :) but I do not read it the same way than you |
| 03:34 | <Hixie> | how do you read it? |
| 03:34 | <Hixie> | i'm curious as to what he meant |
| 03:34 | <Hixie> | i'm confused because he immediately sent another e-mail asking questions |
| 03:34 | <Hixie> | and i don't know if he wants me to reply or not |
| 03:35 | <Hixie> | probably safest not to |
| 03:35 | <karlUshi> | I always try to take a step back for any comments on the mailing-list :/ because I would end up at leaving my job every day ;) |
| 03:35 | <karlUshi> | So the way I read it is |
| 03:35 | <Lachy> | hmm. I read it as shut up too |
| 03:36 | <karlUshi> | "80% is an argument that you always use, I'm not convinced by it, save it for later." |
| 03:36 | <Hixie> | it wasn't intended to be an argument |
| 03:36 | <karlUshi> | ooops |
| 03:36 | <karlUshi> | French |
| 03:37 | <karlUshi> | argument = point of discussion |
| 03:37 | <Hixie> | and i don't understand how saving it for later would make it any more acceptable |
| 03:37 | <karlUshi> | not dispute |
| 03:37 | <Hixie> | right, it wasn't intended to be a point of discussion -- it's just how languages have to be designed |
| 03:37 | <Hixie> | you always have to draw a line as to how much to include |
| 03:37 | <karlUshi> | in your perspective |
| 03:37 | <Hixie> | it's not clear to me that there can be useful perspectives where you don't draw a line |
| 03:39 | <Hixie> | though i'd love to hear of one if there are some |
| 03:39 | <Lachy> | of course the line has to be drawn somewhere, it's just a matter of finding out wheere |
| 03:39 | <karlUshi> | imagine a line based on the inclusion of diversity more than the democracy (democracy being the power of the majority as opposed to the Republics without entering too much into details) |
| 03:39 | <Hixie> | what do you mean by "diversity"? |
| 03:40 | <karlUshi> | we are drifting... but let's say. For example, the republics guarantee the rights of minority because the respect of this diversity is important for the design of a society. |
| 03:40 | <karlUshi> | Democracies at the opposite impose the power of big numbers on minority |
| 03:40 | <karlUshi> | usually a mix of the two is useful. |
| 03:40 | <Hixie> | ok... (looking for how this applies to language design) |
| 03:41 | <karlUshi> | accessibility for example, or internationalisation. |
| 03:42 | <karlUshi> | a feature can be useful for a limited number of persons, but without the feature, the language will not be useful in this precise context. |
| 03:42 | <karlUshi> | Same for the inclusion of some rare characters in unicode. |
| 03:43 | karlUshi | suddenly remembers how much usenet sucks when we could not type accents typing French. |
| 03:44 | <karlUshi> | When some people started to mention that it would be cool to modify the implementation. Some people said it was folly. |
| 03:45 | <Lachy> | karlUshi, I don't understand the point you are trying to make. I thought you were trying to explain a perspective where a line doesn't have to be drawn, but then you gave an example where it was drawn in a different place. |
| 03:46 | <karlUshi> | Lachy: I just said there is not one unique line, I said it was a question of perspective, point of view. :) |
| 03:46 | <karlUshi> | being binary doesn't help |
| 03:47 | <Hixie> | well sure, my point was just that you do have to draw a line |
| 03:47 | <Hixie> | which you seem to agree with, despite saying earlier that that was just "in my perspective" |
| 03:48 | Hixie | is now confused as to both karl's opinion _and_ leif's opinion! |
| 03:49 | <Lachy> | I understand Lief's opinion, he just objects to the 80%-20% argument because he sees it as a way to ignore the minority, particularly on accessibility |
| 03:49 | <Hixie> | ooo, yet another mailing list |
| 03:50 | <karlUshi> | your line is 80/20 rationale. what I said being your perspective :) |
| 03:50 | <Hixie> | Lachy: ah, then he's probably misunderstanding it. i thought i was pretty careful about saying it was 80% of _tables_, not 80% of _users_. |
| 03:50 | <karlUshi> | you decided to draw the line on this principle |
| 03:50 | <Lachy> | Hixie: yes, a lot of people do misunderstand it |
| 03:51 | <karlUshi> | Lachy: because it is often used not only to categorize data, but to remove options to people. |
| 03:51 | <Lachy> | some don't realise that it's not a fixed ratio either, it needs to be evaluated on a case by case basis depending on numerous other factors |
| 03:52 | <Hixie> | what's the status of the design proposals? can i add one? |
| 03:52 | <Hixie> | principles, rather |
| 03:52 | <Lachy> | maybe, what are you thinking of adding? |
| 03:52 | <karlUshi> | So I guess each time we categorize data like this, we have to propose something that will not exclude people falling into the 20%. |
| 03:52 | <Hixie> | people don't fall into percentages |
| 03:53 | <Hixie> | it's 80% of _content_, not 80% of _users_ |
| 03:53 | <Hixie> | Lachy: the line |
| 03:53 | <karlUshi> | Hixie: unfortunately managerial decisions show every days that people fall into percentages. I'm pretty I just have to read Yahoo News right now |
| 03:53 | <karlUshi> | pretty sure |
| 03:54 | <Hixie> | i'm just talking about language design here |
| 03:54 | <Hixie> | nothing else |
| 03:54 | <Hixie> | and in designing languages for the web, we should aim to cater for 100% of users, and n% of use cases, where n < 100. |
| 03:54 | <karlUshi> | yes but people are "humans" not machines. When they read something, they charge it with emotions, etc. The same way you read leif comment with emotions. |
| 03:54 | <Hixie> | lachy: (that's the principle i was going to add) |
| 03:55 | <karlUshi> | so when you say 80% of something, many people will think 20% of people |
| 03:55 | <Hixie> | if people aren't going to read what i write, that's not really my problem |
| 03:55 | <Hixie> | :-) |
| 03:55 | <Hixie> | at least not in this context |
| 03:56 | <karlUshi> | :) unfortunately it is ;) or you have missed everything about social relationship :p |
| 03:59 | <Hixie> | not in this context -- if people are going to ignore what i write and argue against what they think i'm writing, then i'll just ignore them, as they aren't being productive. |
| 03:59 | <Hixie> | no big loss to me, since someone who doesn't slow down long enough to actually read what they're talking about probably isn't going to be slowing down long enough to think about what they're writing either. |
| 04:00 | <Lachy> | exactly! It's more productive to not respond in some cases |
| 04:01 | <Lachy> | and just hope that, eventually, people get the idea that unproductive, emotionally charged responses don't get responses, and then they stop sending them |
| 04:05 | <Lachy> | oh, nice, Jason White gave a usecase for <input usemap> |
| 04:05 | <Lachy> | yet he's ignoring that virtually *nobody* uses it for that purpose anyway |
| 04:15 | <Hixie> | ok, added Baby Steps to the principles |
| 04:19 | <Lachy> | that's quite well written |
| 04:19 | <Hixie> | why thank you |
| 04:19 | <Lachy> | hmm. I wonder if it's at all possible to search for sites that use <input type=image> as a server side image map |
| 04:20 | <Hixie> | as opposed to a button? |
| 04:20 | <Hixie> | i.e. those that care about the .x and .y coordinates? |
| 04:20 | <Lachy> | yes |
| 04:20 | <Lachy> | but there'd be a lot that use type=image, but don't care about x,y |
| 04:20 | <Hixie> | right |
| 04:22 | <Hixie> | i don't understand why people have such a gut reaction to us removing unused or abused features |
| 04:22 | <Hixie> | (like usemap) |
| 04:22 | <Hixie> | (on input) |
| 04:23 | <Hixie> | the philip in http://www.w3.org/mid/46C1DE89.1080305⊙RAU apparently missed the list of URIs that has been cited several times |
| 04:23 | <Lachy> | some people just seem to think the designers of HTML4 had more wisdom then we do, and think they must have included it for a good reason |
| 04:50 | <Lachy> | I think I know how we could search for type=image used as a server side image map... |
| 04:50 | <Hixie> | oh? |
| 04:51 | <Lachy> | my hypothesis is that if the only visible control in a form is type=image (may also include type=hidden or readonly controls), then it's more likely that the site would be interested in the x,y coords |
| 04:52 | Hixie | replies to 58 e-mails about alt="" |
| 04:52 | <Lachy> | also, larger images are more likely to be image maps, than smaller images which are likely to be simple buttons |
| 04:53 | <Lachy> | so searching for pages that meet those criteria should reveal the pages that use it |
| 04:55 | <Lachy> | should I send you a mail about that for you to put on your todo list to do the search? |
| 04:55 | <Hixie> | it would be hard to do in my current framework, because my parser doesn't really do the form association stuff |
| 04:55 | <Lachy> | ok |
| 04:56 | <Hixie> | and i'm not at all convinced that there aren't many pages with large <input type=image>s :-) |
| 04:56 | <Lachy> | yeah, and the only way to know how big the image is, is to retrieve it, which would be quite a slow process |
| 04:57 | <Hixie> | my current framework can't handle looking at external files at all |
| 04:57 | <Hixie> | and making it do so would be far more work than i'm interested in doing |
| 04:58 | <Hixie> | (FAR more work) |
| 04:58 | <Lachy> | eyah |
| 04:58 | <Lachy> | yeah |
| 04:58 | <Hixie> | though height/width might help |
| 05:00 | <Lachy> | oh look, Robert has described hypothetical use cases for <input usemap> |
| 05:00 | <Hixie> | i hope he makes sure to create a wiki page for each of those use cases / problems |
| 05:00 | <Hixie> | so i can make sure to address them |
| 05:01 | <Lachy> | I'll get them started |
| 05:02 | <Hixie> | (don't bother unless the use cases are compelling) |
| 05:02 | <Hixie> | ok, home time. bbl. |
| 05:02 | <Lachy> | ok, fair enough |
| 05:02 | <Lachy> | cya |
| 08:02 | <Lachy> | ah, I'm not looking forward to the backlash against Hixie's comment "I don't really have much time for process" |
| 08:03 | <Lachy> | (although I actually agree with it) |
| 08:06 | <Hixie> | the guy told me i shouldn't have edited the wiki |
| 08:06 | <Hixie> | fuck that |
| 08:06 | <Lachy> | yeah, I know, that's what I thought too |
| 08:06 | <Hixie> | on a more practical note, we need to work out a story for height and width on <img> |
| 08:06 | <Lachy> | what about them? Are they being dropped? |
| 08:07 | <Hixie> | working out what they mean, what their conformance criteria are, etc |
| 08:07 | <Hixie> | i think it was hsivonen who pointed out that it's getting quite common for pages to include extremely high res pictures, and then they are using height=/width= to make them size sanely even in the absence of CSS |
| 08:08 | <Hixie> | i'm a big fan of making sure pages can render sanely even without CSS (or with difference CSS) |
| 08:08 | <Lachy> | BTW, I just reworded cowpaths and reinventing the wheel principles in the wiki too |
| 08:08 | <Hixie> | different, even |
| 08:08 | <Hixie> | but it seems weird for the page to give a particular size |
| 08:08 | <Hixie> | since that's very media-specific |
| 08:08 | <Hixie> | what if my device is bigger? smaller? |
| 08:09 | <Lachy> | some sites that use high res images and then scale them down seem to do so because they don't know how to operate an image editor |
| 08:09 | <Hixie> | when i've done it it's been because i wanted the print version to have access to more pixels |
| 08:09 | <Hixie> | these days it's probably also to let higher res displays have access to more pixels |
| 08:12 | <Hixie> | i guess i can pretend that the height/width are encoding the "real" dimensions of the image in CSS pixels |
| 08:12 | <Lachy> | wouldn't that be a use case for media queries on <img>? |
| 08:12 | <Lachy> | although, I"m not sure how practical that would be to implement, it's what XHTML2 have done, I think |
| 08:13 | <Hixie> | i don't think there's anything wrong with simply providing a very high res image |
| 08:13 | <karlUshi> | Hixie: replied |
| 08:13 | <Hixie> | and expecting the UAs to scale appropriately |
| 08:13 | <Lachy> | ok |
| 08:14 | <Hixie> | karlUshi: thanks |
| 08:15 | <arve__> | <Hixie> i guess i can pretend that the height/width are encoding the "real" dimensions of the image in CSS pixels <-- care to elaborate? |
| 08:15 | <arve__> | It's the word "real" I don't quite get |
| 08:15 | <Hixie> | i think i will pretend that the height/width are encoding the "real" dimensions of the image in CSS pixels, where those dimensions might differ from the actual dimensions of the image because we're expecting UAs to ignore DPI info in the image, but the ratio has to match the real ratio |
| 08:16 | <Hixie> | so if the image is actually 1000dpx by 500dpx, but the author puts width=100 height=50, then that's ok |
| 08:16 | <Hixie> | but if they have an image that's 100dpx by 50dpx and they put height=100 width=100 that's bad |
| 08:16 | <Hixie> | dpx = device pixel |
| 08:16 | <Hixie> | as opposed to CSS pixel |
| 08:16 | <Lachy> | ok, so they have to keep aspect ratios to be conforming |
| 08:17 | <arve__> | what should happen in the latter case, then? |
| 08:17 | <arve__> | I think there are legitimate uses of that for image stretching purposes |
| 08:17 | <Hixie> | arve__: oh i'm not changing anything that happens, it'll stay as per today |
| 08:17 | <Hixie> | arve__: like? |
| 08:18 | <arve__> | for instance, screencaps from some video source may be aspect-ratio-distorted, and you can use width/height to carefully correct it |
| 08:19 | <arve__> | or the other case, which I concede is abuse, to create progress bars and similar using images |
| 08:19 | <Lachy> | arve__: wouldn't it be better to correct such distortions in an image editor, rather than letting browsers scale it? |
| 08:19 | <arve__> | Lachy: surely, but people do stuff like that anyway |
| 08:19 | <Lachy> | browsers tend to scale images quote poorly compared with image editors, which use more complex algorithms |
| 08:19 | <Lachy> | really? Do you have any examples? |
| 08:20 | <Lachy> | but just because people do that, should we allow it? People do a lot of other invalid stuff |
| 08:20 | <arve__> | Lachy: other than "from memory", no? |
| 08:21 | <Lachy> | ok |
| 08:21 | <arve__> | I've distorted aspect ratio on purpose in scripting, though |
| 08:23 | <Lachy> | Hixie: I've found several sites that do actually depend on <input usemap> working, and which break in IE |
| 08:23 | <Lachy> | But all of them would be solved by using <a><img></a> or <img usemap> |
| 08:38 | <Lachy> | Hixie, these sites were listed in your results, but they don't contain <input usemap> http://gctinc.com/index.html http://globulus-ural.com/links.php |
| 09:06 | <Hixie> | virtuelv_: if the images are aspect-ratio distorted (that is, anamorphic) then they should be fixed. there's no reason to have anamorphic static images. |
| 09:07 | <virtuelv_> | Hixie: not that I disagree |
| 09:07 | <Hixie> | Lachy: they might have changed since i looked at them |
| 09:07 | <Lachy> | ok, how long ago did you look at them? |
| 09:07 | <Hixie> | Lachy: and yes, some pages might require <input usemap>, but unless it's a majority... |
| 09:08 | <Lachy> | is it possible that the usemap actually occured on other pages of the site and you listed the wrong page? |
| 09:08 | <Hixie> | no |
| 09:08 | <Hixie> | but the data can be several weeks out of date depending on exactly what i was doing |
| 09:09 | <Lachy> | oh right, since it's from Google's cache |
| 09:22 | <Whiskey_M> | 'lo |
| 09:22 | <Hixie> | hey |
| 09:23 | <hsivonen> | Hixie: do you expect conformance checkers to download images and check if they are fully transparent? |
| 09:24 | <Whiskey_M> | Hixie, just saw the e-mail about web forms, I know it's at the backburner, but (and sorry if I should have RTFMed from the website and missed it), is there any effort in to write a server side forms processor + front end JS for new controls? |
| 09:36 | <Hixie> | hsivonen: i dunno |
| 09:37 | <Hixie> | hsivonen: i'm starting to think that maybe we should have a level of conformance for conformance checkers that does involve checking external resources, because i keep coming across features that would be nice to add that require that |
| 09:41 | <hsivonen> | Hixie: anything else besides transparent images and image dimensions? |
| 09:42 | <hsivonen> | (I'm not yet convinced that tying conformance to external resources is a good idea) |
| 09:43 | Whiskey_M | guesses his was a silly question and goes back to the site to look for details |
| 09:44 | <Lachy> | I found one site that would be a valid use case for <input usemap>, except that due to the way it's implemented in Firefox, it's actually better in browsers that don't support it http://realestate.bannisterdesign.com/grostick.com/search.php |
| 09:45 | <Lachy> | the problem is the <area> links don't submit the coordinates where the user clicked, but the site uses them for the search |
| 09:51 | <Hixie> | hsivonen: type="" on <link> and <script> and <object> matching the resource, all kinds of stuff |
| 09:51 | <Hixie> | hsivonen: not having links to 404s |
| 09:52 | <Hixie> | though i guess that should be a should |
| 09:52 | <Hixie> | i'm not sure how we should phrase it |
| 09:52 | <Hixie> | maybe document conformance and document context conformance or some such |
| 09:53 | <hsivonen> | Hixie: the former is theoretically hard, because GET doesn't allow you to see what someone else GETs |
| 09:54 | <hsivonen> | Hixie: this has already been a problem with krijn's and anne's Apache configs |
| 09:54 | <hsivonen> | which makes it a practical problem, too |
| 09:54 | <Hixie> | well, the same can be said of any URI, even the one the validator gets |
| 09:54 | <Hixie> | unless you're uploading the file |
| 09:54 | <hsivonen> | yes |
| 09:55 | <hsivonen> | but at least when the validator has received something from the main URI, it can say definite things about that stream of bytes plus content-type |
| 09:55 | <Hixie> | it's possible that conformance criteria dependent on external resources are only useful in a dynamic environment, same as conformance criteria that apply to script-driven scenarios |
| 09:55 | <Hixie> | e.g. i don't expect a conformance checker to say if the scripts are conforming, but we have requirements on those already |
| 10:24 | <Whiskey_M> | outside of on-going discussions here does anyone have a link to the usemap thread |
| 10:26 | <zcorpan_> | "If either attribute is specified, the other must be as well" [img width/height] -- why? specifying only one is a good way to get the right ratio, especially when you use percentages |
| 10:29 | <Lachy> | I hope my detailed explanation of the usemap issue I just sent puts a conclusive end to the issue. I've wasted too much time on it already today. :-( |
| 10:44 | <hsivonen> | I wonder if the people talking about "education" have ever done upgrade evangelism on behalf of a browser |
| 10:45 | <hsivonen> | that is, contacted sites to explain that the sites are doing something wrong and need to be fixed in order to work with a new browser |
| 10:52 | <Hixie> | zcorpan_: because the point is to allow for authors to prevent flickering page load, not for them to be able to get the layout they want |
| 10:54 | <zcorpan_> | Hixie: right. i thought percentages were allowed but looking again it seems they aren't. ok |
| 10:57 | <zcorpan_> | Hixie: still, having images with more pixels than the size you want it in order to get better quality on high resolution screens, and not caring about page load flickering, it is easier to just specify one if you're authoring by hand |
| 10:57 | <Hixie> | mmm |
| 10:57 | <Hixie> | fair enough |
| 11:07 | <Hixie> | woah, robert's latest e-mail is weird |
| 11:07 | <Hixie> | he's arguing in favour of <input usemap> in response to an e-mail where Lachy showed that <input usemap> was a net accessibility loss |
| 11:14 | <Hixie> | (and citing accessibility reasons for it) |
| 11:15 | Hixie | decides to let others reply to the bulk of that e-mail, but jumps in on the last paragraph |
| 11:16 | <Hixie> | zcorpan_: spec updated |
| 11:19 | <zcorpan_> | Hixie: looks good |
| 11:29 | <Hixie> | hsivonen: lachlan (writing in nov 2006) said, about src="": |
| 11:29 | <Hixie> | > And, as I mentioned in IRC, I think it should be defined that the value |
| 11:29 | <Hixie> | > should resolve to a valid URI for an image |
| 11:29 | <Hixie> | what should i do? say something like "the resource specified by the src attribute must be [an image]"? |
| 11:29 | <Hixie> | or say nothing? |
| 11:30 | <hsivonen> | hmm. perhaps there indeeds to be a class of tests for external resources |
| 11:30 | <hsivonen> | Hixie: what's an image? |
| 11:31 | <Hixie> | that's a separate rat hole that i don't want to get into tonight |
| 11:31 | <hsivonen> | Hixie: is an application/pdf file an image in the Safari context? |
| 11:31 | <Hixie> | but i'll probably define it as image/* resources that correspond to bitmap data |
| 11:31 | <Hixie> | thus excluding pdfs, svg, and wmfs |
| 11:31 | <hsivonen> | Hixie: what about SVG and Opera? |
| 11:34 | <zcorpan_> | who are we helping by having requirements for what <img src> points to? |
| 11:35 | <Hixie> | the original reason for Lachy's request was to be able to ban <img src=""> (pointing to the HTML file itself) |
| 11:35 | <Hixie> | which it would be vaguely useful to discourage |
| 11:40 | <Hixie> | ok i'm gonna sleep now |
| 11:40 | <Hixie> | more <img> nonsense tomorrow |
| 11:44 | <Lachy> | oh wow, this is going around in circles already: "Unless you can layout what problems you think we will solve by eliminating <input type=image usemap= > from HTML, why are you even posing it." |
| 11:44 | <Lachy> | I thought we already outlined the problems that removing it solves |
| 13:19 | <krijnh> | Ping |
| 13:21 | <Philip`> | Ooh, IE actually parses <blink>...</blink> properly and not like an unrecognised tag |
| 13:22 | <Philip`> | Could be quite useful if someone wanted to add a Backward-LINK element in a future version of HTML, since it already has the correct parsing in IE |
| 13:22 | <takkaria> | heh |
| 13:23 | <zcorpan_> | wonder if the serializing html fragments algorithm should drop PIs on the floor instead of bailing out |
| 13:23 | <zcorpan_> | or emit them as "<?" + target + " " + data + ">" |
| 13:24 | <takkaria> | Philip`: if that's not repurposing of elements, I don't know what is |
| 13:24 | <hsivonen> | Philip`: the new rev=''! :-) |
| 17:31 | <Philip`> | I did http://zaynar.demon.co.uk/misc2/font4.png a while ago with Latin, Latin-Extended, Greek, Cyrillic and Hebrew, and it fits alright, but it doesn't seem very scalable to things like CJK |
| 19:51 | <guardian> | hello |
| 19:52 | <guardian> | please, where could i find the canvas spec ? |
| 19:52 | <guardian> | http://www.whatwg.org/specs/web-apps/current-work/#the-canvas remains silent |
| 19:52 | <gavin_> | "remains silent"? |
| 19:52 | <guardian> | empty page :) |
| 19:52 | <gavin_> | not for me |
| 19:53 | <guardian> | really ? |
| 19:53 | <Philip`> | http://www.whatwg.org/specs/web-apps/current-work/multipage/section-the-canvas.html#the-canvas might work better if your browser doesn't like really big pages |
| 19:53 | <grimboy> | Yeah, works for me too. |
| 19:53 | <guardian> | i just get an html page with an empty body |
| 19:53 | <guardian> | i'm using firefox |
| 19:53 | <Philip`> | (Works for me too :-) ) |
| 19:54 | <Philip`> | (but the multipage one works quicker) |
| 19:54 | <guardian> | duh |
| 19:54 | <guardian> | very strange |
| 19:54 | <guardian> | surprisingly ie7 displays it |
| 19:54 | <guardian> | well thanks for the other link |
| 19:55 | <guardian> | while i'm at it, is there a way i could make a canvas element have the same size as the client area of the browser window ? |
| 19:56 | <Philip`> | Is it adequate to rescale it with CSS, or do you want the bitmap dimensions to change too? |
| 19:58 | <guardian> | well i'm trying to make it have the same size as the client area, then i want to draw something inside |
| 19:59 | <zcorpan_> | btw, do sites rely on image maps being css pixels rather than following the size of the image? i've found 1 site that doesn't work correctly because of this behavior |
| 19:59 | <guardian> | i tried <canvas style="width: 100%"> but firefox keeps giving me a 300px wide canvas |
| 19:59 | <zcorpan_> | http://sodertalje.se/ when you change the text size |
| 19:59 | <zcorpan_> | the banner at the top |
| 20:00 | <Philip`> | guardian: I'd expect that to give a 300x150 bitmap which is rescaled to 100% of the width of whatever |
| 20:00 | <Philip`> | If you want to change the bitmap's size, you need <canvas width=... height=...> |
| 20:00 | <Philip`> | or set canvas.width=... etc with JavaScript |
| 20:00 | <zcorpan_> | or rather, Isac Lagerblad found it |
| 20:01 | <guardian> | you mean drawing happens in the 300x150 space then it's interpolated ? |
| 20:01 | <Philip`> | guardian: Yes (since the bitmap's default size is 300x150 if you don't specify width= anywhere) |
| 20:03 | <guardian> | ok i did not know bitmap's default size was 300x150 |
| 20:03 | <guardian> | i'm new to this |
| 20:03 | <guardian> | (to canvas) |
| 20:04 | <Philip`> | http://www.whatwg.org/specs/web-apps/current-work/multipage/section-the-canvas.html#width3 is where the default sizes are defined |
| 20:04 | <guardian> | gonna try to play a bit with javascript to adjust the size then |
| 20:05 | <Philip`> | It might be useful to know that the canvas gets reset to its default empty state whenever you resize it, which might be a pain if you're making it dynamically resize to fit some other object |
| 20:06 | <guardian> | thx |
| 20:06 | Philip` | wonders [in a completely unrelated direction] if anybody actually likes/uses server-side image maps nowadays |
| 20:07 | <Philip`> | I get the impression that they were much more popular when HTML4 was being developed, so it was more worthwhile trying to handle them accessibly; but now anybody who cares about accessibility won't be using server-side image maps at all |
| 20:11 | <hsivonen> | Philip`: I use quite frequently a service that uses server side image maps without using ismap |
| 20:11 | <hsivonen> | dunno how they do it |
| 20:11 | <hsivonen> | http://aikataulut.ytv.fi/reittiopas/fi/?mm=trip&sm=&m=&adv=&map=1&n=&a=8411&b=8127&c=&keya=Kuusitie%252525252525252010&keyb=Karvaamokuja%25252525252525204&keyc=&an=10&bn=4&cn=&date=20070815&time=2216¶ms=1l3l1l3l2l2ll1l1l1l1l1l0&temp_a=2550006%2C6676237&temp_b=2549061%2C6678598&temp_c=&time=2216&date=20070815&which=&px=2547000&py=6679334&ppx=&ppy=&zx=2545666&zy=6670268&zoom=1&width=600&height=400&route=t0a2550006a6676237t0a0a0t9a183a90t1a35b20t9b |
| 20:11 | <hsivonen> | that's an example |
| 20:13 | <Lachy> | hsivonen: I don't see the example image map |
| 20:13 | <Philip`> | That mostly just says "Valitettavasti paikkaa/reittiä ei voitu näyttää kartalla." which I guess translates to "sorry, there isn't an image map on this page" or something similar :-( |
| 20:14 | <Lachy> | do we have to be logged in or something? Or have some cookie? |
| 20:14 | <hsivonen> | hmm. wfm. cookies or something, I guess |
| 20:15 | <hsivonen> | try http://aikataulut.ytv.fi/reittiopas/en/?keya=pasila&a=17816&an=&bb=17838%3At1a2551837a6673309%3AKamppi%3A740%3A&bn=&keyb=kamppi&hour=22&min=23&vm=1&day=15&month=08&year=2007&va=2&adv= and one of the show route links |
| 20:16 | <Philip`> | It's odd how <a href=.><img usemap=#m ismap></a> in IE appears to send 'click' to the 'a' (in areas not covered by <area>s), but then doesn't actually load the page after you've clicked on it |
| 20:16 | <Lachy> | hsivonen: I clicked show route from that page and got the map |
| 20:17 | <hsivonen> | Lachy: so the map is in practice a server-side image map |
| 20:17 | <hsivonen> | Lachy: but in HTML terms, it is not |
| 20:18 | <Philip`> | That's just an <input type=image> |
| 20:18 | <Lachy> | hsivonen: yes it is. It uses <input type=image src=...>, which sends the x,y coords when click |
| 20:19 | <hsivonen> | whoa. I'm really bad at viewing source then |
| 20:19 | Philip` | used Firebug -> Inspect :-) |
| 20:20 | <Lachy> | checking the HTTP request headers revealed the &x=00&y=00 |
| 20:20 | <Lachy> | XRAY would have helped too |
| 20:26 | <Philip`> | Bah, I can't find anything crazy or inconsistent about how Firefox and Opera handle <input usemap> :-( |
| 20:27 | <Philip`> | Oh, except that Opera applies it to type!=image inputs too |
| 20:27 | <Lachy> | lol |
| 20:28 | <Lachy> | I tested non image inputs yesterday in Firefox hoping to find a bug like that. I should have checked opera |
| 20:38 | <zcorpan_> | Philip`: have you filed a bug report? |
| 20:40 | <Philip`> | Hmm, I'm not actually sure it is a bug |
| 20:41 | <Philip`> | HTML4 appears to say that INPUTs with 'ismap' must have type "image", but it says nothing about usemap |
| 20:42 | <Philip`> | (<input usemap> seems to just be covered by "The MAP element specifies a client-side image map (or other navigation mechanism) that may be associated with another elements (IMG, OBJECT, or INPUT)." which doesn't even make grammatical sense) |
| 20:51 | <Lachy> | if I were to narrow down the selection for the 3 forms task forms members, I'd have to go with at least 1 or 2 (if not 3) of these people: Hixie, Maciej, Anne, Matthew Raymond and maybe Ben Boyle |
| 21:24 | <Philip`> | http://yuiblog.com/blog/2007/08/13/rte-notes/ - "My biggest problem with the native execCommand method (in all browsers) is that the browser doesn?t tell you what it applied the command to. ... The world of JavaScript editors would be so much more civilized if this would happen (hint, hint, nudge, nudge)." |
| 21:24 | <Philip`> | That sounds like a feature request for http://www.whatwg.org/specs/web-apps/current-work/#command |
| 21:27 | <jgraham__> | http://blog.ianbicking.org/2007/08/14/reflection-and-description-of-meaning/ |
| 21:58 | <Hixie> | someone moved universal access to the disputed section |
| 21:58 | <Hixie> | who the hell is disputing universal access |
| 22:02 | <tndH> | a slight disagreement over wording apparently makes it disputed |
| 22:22 | <hsivonen> | some of my photos on flickr from May still lack proper *titles* even... let alone descriptions |
| 22:24 | <Philip`> | Would you mind if the validator said those pages were in error, or would it be a useful reminder that you should go back and fix them some time? |
| 22:25 | <hsivonen> | Philip`: I already stuff approximate filler data in there, so a validator wouldn't know |
| 22:25 | <hsivonen> | (I have a sensible title for each photo set and I duplicate the title of the set on untitled photos) |
| 22:27 | <hsivonen> | I don't expect to give proper titles to all my photos in my lifetime |
| 22:27 | <hsivonen> | I don't expect to describe them all, either. |
| 22:31 | <Philip`> | If there hadn't been a decade of alt being required and everybody telling everybody they must use it, would you (or, more particularly, other people) ever put in alt descriptions of images even when they were appropriate? |
| 22:31 | <Philip`> | (I don't know the answer to that, so I can only guess) |
| 22:32 | <hsivonen> | Philip`: on flickr photos, definitely no. on diagrams in my thesis, probably |
| 22:32 | <hsivonen> | Philip`: it between, hard to say |
| 22:34 | <Philip`> | I'd probably guess that if validators couldn't complain about missing alts, and if the conformance rules for it were as complicated as they are now (i.e. you must use alt except when it's too hard and then you mustn't use it) so you couldn't easily explain it to someone, there would be less usage of alt text (both correct and incorrect) |
| 22:56 | <zcorpan_> | http://simon.html5.org/temp/zon/001.htm -- the framework should be pretty good now, so now i can start actually writing tests using it... :) |
| 23:15 | jgraham__ | is supremely unconcerned with whether his flickr page validates |
| 23:17 | <jgraham> | But perhaps I should be. I mean I have never considered it as an issue when choosing where to put my photos |
| 23:17 | <jgraham> | I guess valid markup would be easier to scrape |
| 23:18 | <jgraham> | if I wanted to retrieve data and the API was taken away |
| 23:18 | <Hixie> | man if http://my.opera.com/oedipus/albums/ is supposed to be an example of good alt text, my standards are too high |
| 23:18 | <Hixie> | (i tried switching to no CSS and turned off images and that page and its subpages became unusable) |
| 23:24 | <jgraham> | Yeah, those images seem to have no useful alt text |
| 23:26 | <Philip`> | Hixie: Maybe it indicates that your perspective is not the same as that of the target audience for alt text |
| 23:29 | <Hixie> | Philip`: i'd love to see that perspective |
| 23:40 | <Hixie> | Philip`: so... i'm basically ignorant of the statistics stuff, which i should have learnt at university but it seems i've forgotten all of it |
| 23:40 | <Hixie> | Philip`: how can i determine what the accuracy of my numbers is? |
| 23:40 | <Hixie> | that is, how can i work out the error margin? |
| 23:40 | <Hixie> | the way that you do |
| 23:44 | <Hixie> | (feel free to reply here, i'll be back online in about 25 minutes) |
| 23:44 | <Philip`> | I'm basically ignorant of it too, but I think I can remember just enough to find the parts of Wikipedia that look relevant and non-vandalised :-) |
| 23:46 | <Philip`> | When looking at a random sample of pages and getting a binary value for each (e.g. page does/doesn't have <blink>), the measured mean should follow a binomial distribution (i.e. each value is 'yes' with a probability p, independent of all other values) |
| 23:47 | <Philip`> | and, as in <http://en.wikipedia.org/wiki/Binomial_distribution#Normal_approximation>;, that can be approximated as a normal distribution with mean n*p and variance n*p*(1-p) (as long as n*p and n*(1-p) are large enough, like about 10) |
| 23:47 | <Philip`> | and then standard deviation = sqrt(variance) = sqrt(n*p*(1-p)) (by definition) |
| 23:48 | <Philip`> | and, as in <http://en.wikipedia.org/wiki/Normal_distribution#Standard_deviation>;, about 95% of values are within two standard deviations of the mean |
| 23:50 | <Philip`> | so there is a 95% probability that the real population mean (from which the sample was taken) is within 2 std devs of the measured sample mean |
| 23:51 | <Philip`> | p is the real population mean, which you can't know since you're just looking at a sample, but there's probably some theory somewhere that says it's valid to do this with p = the measured sample mean |
| 23:52 | <Philip`> | which means you can count n (sample size), calculate p (proportion of sample which were true), then with 95% confidence the real population mean is n*p +/- 2*sqrt(n*p*(1-p)) |
| 23:53 | <Philip`> | But I wouldn't mind checking with someone who really knows how to do statistics to make sure that's all valid and not just cheating :-) |
| 23:57 | <Philip`> | (At least that final equation has roughly the right behaviour - the proportional error decreases with sqrt(n), so using a million times larger sample only gives an extra three decimal places of accuracy) |
| 23:57 | <Philip`> | (which is why I'm happy looking at eight thousand web pages instead of eight billion :-) ) |
| 23:58 | <Philip`> | (...at least for things which are >1%, because otherwise n*p gets really small and the data becomes useless if the sample isn't huge) |
