| 07:57 | <hgl> | is it ok to ask es6 questions here? |
| 09:11 | <MikeSmith> | hgl: sure, some tc39 members are around here at times. as well as implementors some |
| 09:14 | <hgl> | @MikeSmith, thank you. after some googling. i think i got the answer. but will ask here next time i got a new question. :) |
| 12:47 | <annevk> | mathiasbynens: note that https://tools.ietf.org/html/rfc3629#section-4 does not allow six bytes either |
| 12:48 | <mathiasbynens> | annevk: hah, may want to leave that as a comment :) |
| 12:49 | <annevk> | sure |
| 13:42 | <arv> | iamstef: I'm interested. I think Guy Bedford has something that works using Traceur too. |
| 14:10 | <allam2002> | hii everyone |
| 14:11 | <allam2002> | i have a problem and i need help can anyone help me?? |
| 14:13 | <jgraham> | I would say that strongly depends on the nature of your problem. |
| 14:14 | <allam2002> | i got an error on my website that say : Bad value uri-translation for attribute name on element meta: Keyword uri-translation is not registered |
| 14:15 | <jgraham> | Ah, then yes we can help |
| 14:15 | <allam2002> | and when i tried to register on WHATWG wiki they told me to ask one one of these permanent autoconfirmed members |
| 14:16 | <jgraham> | MikeSmith might be a good person to look at this if he's awake |
| 14:17 | <allam2002> | this element meta belong to gtranslate website an online traslator |
| 14:18 | <jgraham> | Is there a page somewhere describing how this meta value works? |
| 14:20 | <allam2002> | sorry but i asked them in there fourm site and they replyed that i have to register it myself |
| 14:21 | <allam2002> | and i really have no experience in such thing |
| 14:22 | <caitp> | http://gtranslate.net/features this thing? |
| 14:22 | <allam2002> | yes |
| 14:23 | <allam2002> | they gave me a meta line to put in my website which include this element |
| 14:23 | <jgraham> | So how do you know how to write this kind of element? |
| 14:24 | <jgraham> | Oh they provide one-off examples? There isn't any documentation? |
| 14:24 | <jgraham> | Also I don't really understand what's consuming this |
| 14:25 | <allam2002> | To enable URL translation feature for your website you need to add the following code into the head tag of your pages. |
| 14:25 | <allam2002> | <meta name="uri-translation" content="on" /> |
| 14:26 | <allam2002> | that is why i need help from a pro like you |
| 14:27 | <caitp> | it looks like you register your site with them, they host a database of translated text and crowdsource the translations using one of those cheap labor services |
| 14:27 | <caitp> | and there's probably some code on the client which requests the translated text |
| 14:27 | <allam2002> | that is right |
| 14:29 | <allam2002> | they translate my website as subdomain |
| 14:30 | <allam2002> | http://gtranslate.net/docs/58-gtranslate-enterprise-documentation |
| 14:35 | <zcorpan_> | hsivonen: SELECT url,body FROM [httparchive:runs.2014_08_15_requests_body] WHERE REGEXP_MATCH(LOWER(mimeType), r"(text/html|application/xhtml\+xml);\s*charset\s*=\s*[\"']?utf-16") gives zero matches |
| 14:36 | <zcorpan_> | hsivonen: http://bigqueri.es/t/analyzing-html-css-and-javascript-response-bodies/442 |
| 14:36 | <zcorpan_> | hsivonen: webdevdata also zero |
| 14:38 | <allam2002> | my website is : emperorsource.com |
| 14:53 | <annevk> | allam2002: I can get you an account on the WHATWG Wiki so you can register this <meta> value |
| 14:54 | <annevk> | allam2002: however, ideally that service registers the values they rely upon themselves, as well as the documentation they have for them |
| 14:54 | <annevk> | allam2002: if you still want an account I need a desired username and email address |
| 14:54 | <allam2002> | username : allam2002 |
| 14:55 | <allam2002> | email : allam2002⊙gc |
| 14:55 | <allam2002> | thank you very much annevk |
| 14:56 | <annevk> | allam2002: you should have mail |
| 14:57 | <allam2002> | i did thank you |
| 14:57 | <allam2002> | i will try to do it my self |
| 16:04 | <zcorpan_> | Hixie: the spec's IDLs don't have links to the obsolete section. known? |
| 16:14 | <annevk> | JakeA: yeah, as I said in that bug, either SVG or DOM should change |
| 16:15 | <annevk> | JakeA: we haven't really moved much stuff to DOM yet, e.g. id="" and class="" are still local in most implementations I think |
| 16:16 | <JakeA> | annevk: for some reason I thought SVG extended HTMLElement when it was inlined in HTML, but yeah that wouldn't work |
| 16:17 | <annevk> | That sounds pretty magical |
| 16:17 | <JakeA> | yeah, doesn't make sense |
| 17:11 | <iamstef> | arv: cool, hopefully later this month or early in the near year i will be working to upgrade our module transpiler. Lets see if recast as made any improvements. |
| 18:26 | <Hixie> | zcorpan: odd... |
| 18:32 | <hsivonen> | zcorpan: I guess we should have some telemetry for UTF-16. Lore has it that UTF-16 exists in Japan. |
| 18:33 | <zcorpan> | speaking of stats, there is surprisingly much of <menu type=context> in httparchive |
| 18:33 | <Hixie> | zcorpan: wait, the IDL thing works for me |
| 18:34 | <Hixie> | zcorpan: can you be more specific? |
| 18:34 | <zcorpan> | Hixie: https://html.spec.whatwg.org/multipage/semantics.html#the-html-element |
| 18:34 | <Hixie> | oh interesting |
| 18:35 | <zcorpan> | Hixie: i see it works on single-page |
| 18:35 | <Hixie> | seems it only works in single-page mode |
| 18:35 | <Hixie> | odd |
| 18:35 | <Hixie> | file a bug? |
| 18:35 | <zcorpan> | ok |
| 18:35 | <Hixie> | thanks |
| 18:36 | <zcorpan> | https://www.w3.org/Bugs/Public/show_bug.cgi?id=27539 |
| 19:01 | <Hixie> | anyone here understand Web Animation well enough to explain how "pause" works? |
| 19:06 | <JonathanNeal> | Does someone here understand the color transform matrix? I would love to understand it better. |
| 19:07 | <JonathanNeal> | e.g. http://www.svgbasics.com/filters4.html |
| 19:23 | <ondras> | well apparently you use the color matrix to perform certain color operations |
| 19:23 | <ondras> | those that can be represented as linear combination of color channels + constant offset |
| 19:24 | <ondras> | so you have a color matrix "desaturate"; by multiplying your color (expressed as a 4-item vector) with this matrix, you create a new - desaturated - color |
| 19:24 | <ondras> | I am not sure, though, how can one characterize the set of (color) operations that can be expressed in this way. |
| 19:28 | <JonathanNeal> | ondras: and I’ve seen that color transform matrix can be used to grayscale an image, but I’d love to know how they could be used to tint an image (like sepia, but with another color) |
| 19:31 | <ondras> | JonathanNeal: the matrix modifies each color channel by an arbitrary amount of other channels and a constant |
| 19:33 | <JonathanNeal> | So, if I wanted to have black be black, white be white, and everything inbetween be the grascale equivelent to some color - would i only modify columns of colors or rows of colors or both? |
| 19:35 | <ondras> | I am not sure. If you want black to stay black, the constant offset must by 0 (so the last column is all zeros) |
| 19:36 | <ondras> | but I would say that it is achievable by picking proper values, probably for all 16 coefficients |
| 19:38 | <JonathanNeal> | ondras: how do i figure out a coefficient? |
| 19:41 | <JonathanNeal> | @ondras and what is the last column. It goes Red, Green, Blue, Alpha, then … ? |
| 19:42 | <zcorpan> | who owns whattf.org? |
| 19:42 | <ondras> | JonathanNeal: the last column is the constant offset |
| 19:42 | <ondras> | JonathanNeal: so basically every row (there are 4 of them) tells you how to compute a channel |
| 19:42 | <ondras> | first row computes red |
| 19:42 | <ondras> | second green |
| 19:42 | <ondras> | etc |
| 19:43 | <ondras> | so the first row used for red computation |
| 19:43 | <ondras> | has five coefficients |
| 19:43 | <ondras> | you multiply r,g,b,a with four of them and add the fifth |
| 19:43 | <ondras> | (the fifth one is zero in your case) |
| 19:45 | <JonathanNeal> | so, the columns don’t actually represent r, g, b, a, and something else, ondras |
| 19:45 | <JonathanNeal> | and the rows DO represent r, g, b, a, though, right? |
| 19:45 | <caitp> | zcorpan: eetemad⊙pe, based on dns records? |
| 19:46 | <zcorpan> | caitp: yeah |
| 19:46 | <zcorpan> | pinged fantasai in #css |
| 19:46 | <ondras> | JonathanNeal: they represent coefficients. existing r,g,b,a values are multiplied with them |
| 19:46 | <ondras> | to produce new r,g,b,a vlaues |
| 19:46 | <ondras> | *values |
| 19:46 | <caitp> | it is a pretty good domain name though |
| 19:46 | <ondras> | JonathanNeal: http://stackoverflow.com/questions/21977929/match-colors-in-fecolormatrix-filter |
| 19:47 | <ondras> | this might provide some insight |
| 19:54 | <JonathanNeal> | @ondras thanks, now i have a general idea of what that means. |
| 19:55 | <rubys> | annevk: there has been a suggestion to capture some of the rationale for the URL Standard in the form on an IETF Internet Draft "Problem Statement". Is this something you would be interested in contributing to? See: http://lists.w3.org/Archives/Public/public-ietf-w3c/2014Dec/0025.html |
| 19:58 | <JonathanNeal> | @ondras how could anyone ever know how to make an image grayscale without already knowing the color there? |
| 20:02 | <ondras> | JonathanNeal: well one of the most straightforward ways of creating grayscale is doing (r+g+b)/3 |
| 20:03 | <JonathanNeal> | but if you don’t know the rgba of the pixel, how would you know? |
| 20:03 | <ondras> | JonathanNeal: so basically when you put .333 as coefficients in the upper-left 3x3 submatrix |
| 20:03 | <ondras> | you know the rgba of the pixel. |
| 20:03 | <ondras> | applying the matrix means "take rgba of a pixel, multiply it with a matrix, you now have rgba of the resulting pixel" |
| 20:05 | <JonathanNeal> | okay, so if my color was red, the matrix would apply .333 * (1), .333 * (0), *0.333 * (0)? |
| 20:05 | <ondras> | yes |
| 20:05 | <ondras> | resulting red channel will be .333 |
| 20:05 | <JonathanNeal> | Ah, and that would be for the row of Red, so the resulting color for red would 0.333 |
| 20:06 | <ondras> | right. |
| 20:06 | <ondras> | the same would be for green and blue rows as well |
| 20:06 | <ondras> | => result color is .33,.33,.33 => gray |
| 20:06 | JonathanNeal | figures out yellow, just to be sure. |
| 20:07 | <ondras> | yellow would be .66,.66,.66 - twice the lighness |
| 20:07 | <ondras> | *lightness |
| 20:09 | <JonathanNeal> | Is that what the fifth column represents? Lightness? |
| 20:09 | <ondras> | no |
| 20:10 | <ondras> | the fifth column represents constant shift, independently on current values |
| 20:10 | <JonathanNeal> | got it |
| 20:10 | <ondras> | so for instance, a matrix with all zeroes but ones in the last column |
| 20:10 | <ondras> | would create white |
| 20:10 | <ondras> | from any input color |
| 20:11 | <JonathanNeal> | Right. |
| 20:11 | <JonathanNeal> | So, if I make the alpha column 0 every time, why isn’t it black? |
| 20:12 | <JonathanNeal> | i guess it doesn’t matter, as long as the alpha / alpha cell is 1? |
| 20:13 | <ondras> | I am not sure about the alpha values. I suppose 1 means fully transparent? |
| 20:13 | <JonathanNeal> | @ondras so if i want something like greyscale, but more like colorscale, would you help me understand what that means I need to do? I might get it better this time. |
| 20:13 | <ondras> | alpha column 0 effecticely means "do not adjust color channels by alpha channel" |
| 20:16 | <ondras> | JonathanNeal: as far as "colorscale" goes, my highly experimental guess would be "take the target color, stick it into the last column multiplied with 0.5; put .1667 into 3x3 upper left submatrix" |
| 20:16 | <ondras> | probably. |
| 20:16 | <JonathanNeal> | What would happen if I made all the colors alpha 1 instead of 0? |
| 20:16 | <TabAtkins> | Alpha 0 is fully transparent. Alpha 1 is fully opaque. |
| 20:16 | <ondras> | I really have no idea. I know how colors work and I know how matrix multiplication works, but this requires some experience and imagination. |
| 20:16 | <ondras> | thanks |
| 20:17 | <ondras> | JonathanNeal: putting anything else than 0 into the alpha column (with the exception of the alpha row) has little sense in your case |
| 20:17 | <TabAtkins> | JonathanNeal: Note that taking the sum of the channels multiplied by 1/3 produces *a* grayscale, but not a good one. It doesn't match up well to our vision's notion of brightness. |
| 20:17 | <JonathanNeal> | @TabAtkins but if you produce a red with 1 (red) 0 (green) 0 (blue) 0 (alpha) 0 (shift) the color still shows up. |
| 20:17 | <ondras> | you probably do not want to adjust your color channels based on the alpha |
| 20:17 | <TabAtkins> | JonathanNeal: point me to the filter? |
| 20:18 | <ondras> | JonathanNeal: you only described the first row coefficients; those influence the target red channel |
| 20:18 | <JonathanNeal> | @TabAtkins is that why I see a pattern online a lot that goes something like .393, .769, .189, etc? |
| 20:18 | <TabAtkins> | JonathanNeal: Yup, exactly. |
| 20:18 | <ondras> | JonathanNeal: if you want to tackle alpha, you want to adjust fourth row. |
| 20:18 | <TabAtkins> | Most of the brightness of a color comes from the green channel. |
| 20:18 | <ondras> | right, we are more sensitive to green |
| 20:18 | <ondras> | you just want those numbers to sum up to 1 |
| 20:19 | <ondras> | on the other hand, one way of desaturating is (max(r,g,b)+min(r,g,b))/2 |
| 20:19 | <ondras> | this is not achievable using the matrix |
| 20:19 | <JonathanNeal> | okay, that’s really cool. |
| 20:19 | <ondras> | relevant: http://docs.gimp.org/en/gimp-tool-desaturate.html |
| 20:19 | <JonathanNeal> | Both what you ondras and TabAtkins have said. It’s really awesome. |
| 20:20 | <TabAtkins> | ondras: I mean, if you're doing it right you don't work in RGB space at all; it's a terrible colorspace with no redeeming qualities besides "it's what (some) monitors use, if you don't have a wide-gamut monitor or a 4-color monitor or whatever". |
| 20:21 | <TabAtkins> | You should be doing all transformations in Lab space, which actually corresponds to human vision. |
| 20:21 | <TabAtkins> | And then converting back to RGB for display purposes only. |
| 20:21 | <TabAtkins> | But that's not doable with a filter. ^_^ |
| 20:21 | <ondras> | :) |
| 20:21 | <JonathanNeal> | @TabAtkins can you do that with SVGs? |
| 20:21 | <TabAtkins> | JonathanNeal: Nope. You gotta do it yourself in script. |
| 20:21 | <Philip`_> | You can convert to YUV with a matrix, which isn't quite as terrible as RGB :-) |
| 20:22 | <ondras> | never the less, I am actually quite interested in describing those color transformation that can be represented by this particular matrix multiplication (4x5). |
| 20:22 | <ondras> | I heard of this for the first time today. |
| 20:23 | <ondras> | game idea - replace all .xyzw in shader code with .rgba and vice versa |
| 20:24 | <Philip`_> | .stpq is more fun |
| 20:25 | <JonathanNeal> | @ondras forgetting about the badness of rob, you said i would want to use only the shift column? |
| 20:25 | <JonathanNeal> | *badness of rgb, blasted auto-correct. |
| 20:25 | <JonathanNeal> | *apologies to rob |
| 20:25 | <ondras> | JonathanNeal: I am not sure; where? |
| 20:25 | <TabAtkins> | Philip`_: YUV is still based on particular colorants, so you can't do arbitrary work on it and expect consistent results across displays. But it is indeed somewhat better shaped, particularly since it has an accurate lightness channel. |
| 20:26 | <JonathanNeal> | when colorscaling an image, i would try to adjust which columns? |
| 20:27 | <ondras> | JonathanNeal: the upperleft 3x3 (fill with .167) and the last column (fill with targetcolor/2) |
| 20:27 | <ondras> | probably. |
| 20:27 | Philip`_ | has been doing a lot of work around cameras, where almost everything is YUV |
| 20:27 | <ondras> | I am not an expert in this area :) |
| 20:27 | <Philip`_> | (except for the bits that are Bayer RGB, which is much nastier) |
| 20:28 | <Ms2ger> | Philip`_, want to do work around canvas? ;) |
| 20:29 | <Philip`_> | Ms2ger: Nah, canvas is so 2006 |
| 20:32 | <ondras> | so 2006, yet still too sexy for my exployer, who needs to support IE8 .-) |
| 20:32 | <caitp> | you know |
| 20:33 | <caitp> | if school districts in north america had a bit more money |
| 20:33 | <caitp> | smh :c |
| 20:34 | <caitp> | school districts <and other government funded enterprises> <and also privately funded enterprises that bought their lasts truckload of dell machines in 2003> |
| 20:35 | <ondras> | yeah well this is not related to the u.s. of a. |
| 20:35 | <ondras> | we are located in .cz, focusing solely on our people |
| 20:35 | <ondras> | our people with windows xp :-) |
| 20:58 | <zcorpan> | i don't understand why <menu> has so much adoption |
| 20:59 | boogyman | doesn't understand many things… most of them are not relevant here. |
| 21:09 | <annevk> | rubys: probably not (what helps btw is when I'm copied on an email) |
| 21:11 | <rubys> | annevk: ok. I'll keep you posted as this evolves (should it in fact do so, I have my doubts) in case you want to participate. |
| 21:12 | <annevk> | rubys: sgtm, if the IETF wants a problem statement they should look at their (private) email archive; this has been done to death |
| 21:13 | <Domenic> | annevk: an expansion of the parenthetical "(E.g. spaces, other "illegal" code points, query encoding, equality, canonicalization, are all concepts not entirely shared, or defined.)" somewhere, e.g. a markdown document in whatwg/url, would be pretty interesting reading I think |
| 21:13 | <Domenic> | although I guess http://w3cmemes.tumblr.com/post/98762809962/hitler-reacts-to-the-html-url-normative-reference gave me some ideas |
| 21:14 | <rubys> | on the topic of canonicalization, the URL Standard doesn't define such outright. And to take one example: given that lowercase percent encoded characters are retained as is, the parsing alone doesn't canonicalize URLs. |
| 21:15 | <Domenic> | I think it requires parsing + serializing? |
| 21:15 | <annevk> | rubys: it does canonicalize, just a different kind |
| 21:15 | <annevk> | rubys: whether there should be some form of normalization exposed browsers don't perform right now is a distinct question |
| 21:16 | <rubys> | annevk: are http://example.com/%7Esmith/ http://example.com/%7esmith/ supposed to be treated the same? See http://intertwingly.net/blog/2004/07/31/URI-Equivalence from over a decade ago. |
| 21:16 | <annevk> | rubys: nope |
| 21:16 | <annevk> | rubys: see https://annevankesteren.nl/2012/09/url-equivalence where I replied |
| 21:18 | <rubys> | never saw that before. Even though somebody took the time to add a link to that as a comment in my weblog. |
| 21:19 | <annevk> | I think I might have tried adding a link back then but your comment system became somewhat hostile over the years... |
| 21:20 | <rubys> | annevk: sorry about that. I really need to rewrite my weblog software. |
| 21:21 | <annevk> | maintaining small pieces of software is hard :( |
| 21:22 | <rubys> | Meanwhile, I'm having fun rewriting my url-testresults page to have all of the analysis done on the client instead. This should make it easier to present the information in clearer ways. I think Domenic will approve. :-) |
| 21:24 | <Domenic> | :) |
| 21:26 | <annevk> | Domenic: it seems more meaningful to list differences with 3986 once implementations have converged |
| 21:27 | <Domenic> | annevk: yeah, very true, just would be interesting to kind of educate people who say "who needs more than 3986" the extent to which 3986 underspecifies |
| 21:28 | <rubys> | is is not just underspecifies. All but the last line of uri/url equivalence is direct from the RFC |
| 21:28 | <annevk> | It is not shy on details to my knowledge, it's just not what browsers implement and a significant part of the web uses |
| 21:29 | <annevk> | gotta go, nn all |
| 21:30 | <rubys> | in any case, I intend to document this rationale. Even if annevk doesn't directly contribute text, he is an invaluable resource on this subject. Simply by providing a link (such as what he did a few minutes ago) is very helpful. |
| 22:00 | <Hixie> | zcorpan: there's a flaw in your study |
| 22:00 | <Hixie> | zcorpan: <menuitem> <div></div> </menuitem> wouldn't get counted |
| 22:04 | <zcorpan> | Hixie: that's right. i expected people to just put text there. but i can run a new search if you can come up with a regexp that doesn't match <menuitem></menuitem> <menuitem></menuitem> |
| 22:05 | <Hixie> | i think the examples that were brought up all have markup there |
| 22:06 | <Hixie> | (i dunno what they wanted to have happen) |
| 22:06 | <Hixie> | btw, thanks for doign this |
| 22:06 | <Hixie> | it's a huge help |
| 22:48 | <zcorpan> | Hixie: np |
| 22:50 | <TabAtkins> | zcorpan: Just use a non-greedy search, no? /<menuitem[^>]*>.*?</menuitem>/ |
| 22:51 | <zcorpan> | TabAtkins: i guess that could have worked, yeah |
| 22:53 | <Domenic> | pretty excited about menus actually being usable |
| 22:53 | <Domenic> | years later |
| 22:53 | <Domenic> | maybe one day we can hope for sortable tables |
| 22:53 | <Domenic> | or date inputs |
| 22:58 | <caitp> | are date inputs not currently usable? |
| 23:01 | <caitp> | seems pretty good on my iphone |
| 23:01 | <caitp> | shrug :< |
| 23:02 | <caitp> | i guess it would be easier to find my birthdate if i were born in the 90s, but other than that, it works okay |
| 23:08 | <Domenic> | i was thinking desktop :P |
| 23:08 | <Domenic> | in particular firefox |
| 23:08 | <TabAtkins> | What, where it's not implemented? |
| 23:10 | <Domenic> | yep, so you can't reliably use date inputs on a desktop site without a polyfill |
| 23:10 | <TabAtkins> | Ugh, the Lollipop date picker is super-jank when you click on a date. |
| 23:10 | <TabAtkins> | Like a half-second delay between clicking and it updating. |
| 23:11 | <TabAtkins> | How... how does that even happen. |
| 23:11 | <TabAtkins> | And day of the week is cut off. |
| 23:11 | <TabAtkins> | brb, filing bugs |
| 23:13 | <zcorpan> | onclick="setTimeout(update, 250 + (Math.random() * 250))" |
| 23:13 | <TabAtkins> | It certainly feels like that. |