| 00:25 | <JonathanNeal> | erlehmann: wikipedia uses a background image for the effect, e.g. http://en.wikipedia.org/wiki/Help:Table#Nowrap |
| 00:26 | <erlehmann> | JonathanNeal i see. bloat! |
| 00:30 | <JonathanNeal> | erlehmann: what kind of system are you on? How does http://sandbox.thewikies.com/table-sorting/ look? (the buttons don’t do anything, currently) |
| 00:31 | <erlehmann> | JonathanNeal GNU/Linux, using the conkeror web browser (mozilla with emacs-like key-bindings) i also have chromium here for testing |
| 00:32 | <erlehmann> | the thing you linked looks ugly. i'll make a screenshot |
| 00:32 | <JonathanNeal> | Thanks. |
| 00:35 | <erlehmann> | data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAG0AAAA6AQMAAACAvUVaAAAABlBMVEX///8AAABVwtN+AAAAjUlEQVQokWNgGAUo4D+DDBKPsZmBB8ywcQCRzIxQLs8BEMkO5z4AcxmkeZ7b/6tn4IVyDXuSJY4ZNvB+gHANzoC5/BAuI4z7A2IyCpexybAnHcz9A7bgj7yc/d731Qfs/6A4lv8fClcelUsJQDOZH9VedjTuDxQuMxr3AwqXEY37ANViNO4BVK4Dw5AEAMi9LR5r3QUHAAAAAElFTkSuQmCC |
| 00:35 | <erlehmann> | JonathanNeal this is how it looks like |
| 00:36 | <erlehmann> | looking at the CSS, i can see why! |
| 00:37 | <JonathanNeal> | Yea, that’s ugly, all right. I think I’ll need data uri svgs. Some browsers won’t allow such small font sizes. |
| 00:37 | <erlehmann> | ? |
| 00:37 | <erlehmann> | then do not make it small |
| 00:37 | <erlehmann> | if the browser does not allow small fonts that is for a reason |
| 00:38 | <erlehmann> | i, for example, have a problem with low-contrast imagery |
| 00:38 | <JonathanNeal> | Right, but we’re taking advantage of fonts to create controls. So, just don’t do that, use an SVG. |
| 00:38 | <erlehmann> | to have a smaller control? |
| 00:38 | <erlehmann> | let me look at the styling issue |
| 00:38 | <erlehmann> | wait |
| 00:38 | <JonathanNeal> | Yes. They don’t need to be that large. They need to look like what we see on a select box. |
| 00:39 | <erlehmann> | the select box respects my font size |
| 00:39 | <erlehmann> | and has large arrows |
| 00:39 | <erlehmann> | let me try something |
| 00:40 | <JonathanNeal> | Sure, well, show me a screenshot. I was going to use an SVG and size it to be 1em, respecting the font size of your tables. |
| 00:40 | <JonathanNeal> | While copying the usecase you shared with me, wikipedia. |
| 00:42 | <erlehmann> | table[sortable] th::after { content: '▲▼'; cursor: pointer; display: inline-block; padding: 0 0.5em; } |
| 00:42 | <erlehmann> | works for me |
| 00:42 | <erlehmann> | putting the arrows beside each other |
| 00:43 | <erlehmann> | instead of over each other |
| 00:43 | <erlehmann> | and not position absolute |
| 00:43 | <erlehmann> | but flow! |
| 00:43 | <erlehmann> | so they flow, even in RTL languages |
| 00:43 | <erlehmann> | wait i provide screenshot |
| 00:44 | <caitp> | I'm not sure the spec actually prescribes a UI for sorting columns (although there's no reason not to I guess), a polyfill would only deal with what is actually outlined in the spec, one imagines |
| 00:45 | <erlehmann> | caitp to sort a table you need some UI |
| 00:45 | <caitp> | to do it from user interactions yes, but the user interactions aren't specified |
| 00:45 | <caitp> | they specify an API for sorting tables |
| 00:45 | <erlehmann> | yeah, so take any interactions. like the one with the arrows! |
| 00:46 | <erlehmann> | (which is used pretty often, i guess) |
| 00:47 | <erlehmann> | data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAQQAAAAUAQMAAABRZ9FmAAAABlBMVEX///8AAABVwtN+AAAA4klEQVQoFc3BsU3DQBSA4f+eDQkV0CAQEfYIDIDkjMAIHgOJiDwEE9FHug463NKdlCJ0uQIpLiybnA8SJQMgvo9/o2PEjswlylqiBGKe8QQ3FDXBqbYERyX51CPyQFRBTXBtiZQxHknZcPSUjRMLQnfJominrHl683K06N7dFygOgTOdfLy9KHiipPKTi9UtJUMHAt2YH47ok8BYAxbB8KsmyomU1IM0bFkCJTK5gCLto7EpvYZelZ+jwIFFataa7KqYLe9tQUIve5q/rmbLu2M3aA/ZYdgz6ASErY49Q1r+yDd9/kX9FVlNvQAAAABJRU5ErkJggg== |
| 00:47 | <erlehmann> | JonathanNeal does that look bad? |
| 00:47 | <erlehmann> | (apart from this being a 1-bit picture) |
| 00:48 | <erlehmann> | the arrows just flow |
| 00:49 | <JonathanNeal> | erlehmann: no it doesn’t look bad, but most of this polyfill requires IE9+ anyway, so I stand by things being better with a polyfill. Hacking text to look like a control is … hacky. |
| 00:49 | <erlehmann> | JonathanNeal „stand by things being better“ means what exactly? more complexity? |
| 00:50 | <erlehmann> | some text is there to provide controls. |
| 00:51 | <erlehmann> | JonathanNeal the unicode triangles will certainly provide a) smaller code size b) less complexity c) more compatibility d) “native feeling” since they are in the font the page is in |
| 00:52 | <erlehmann> | JonathanNeal if you license your polyfill permissively i can always rip out the parts i don't like, so please do it. ;> |
| 00:53 | <erlehmann> | JonathanNeal meanwhile can you see the following glyphs? 🗔 🗕 🗖 🗗 🗘 🗙 🗚 🗛 |
| 00:55 | <erlehmann> | Or these? 🔀 🔁 🔂 🔃 🔄 |
| 00:55 | <erlehmann> | maybe i am mistaken, but i think they are intented as labels for controls! |
| 00:55 | <JonathanNeal> | erlehmann: i saw none of the first glyphs you posted. |
| 00:56 | <erlehmann> | JonathanNeal which platform are you on? |
| 00:56 | <JonathanNeal> | Just a bunch of boxes. |
| 00:56 | <JonathanNeal> | Mac. |
| 00:56 | <erlehmann> | okay, then you don't have up-to-date unicode coverage. |
| 00:56 | <erlehmann> | just look at this picture http://daten.dieweltistgarnichtso.net/pics/icons/unifont-symbols-emoji.png |
| 00:56 | <erlehmann> | i am drawing characters for GNU unifont |
| 00:57 | <erlehmann> | and there are quite some who are clearly meant to be part of controls |
| 00:57 | <erlehmann> | i posted the window bar maximize, minimize, restore icons |
| 00:57 | <erlehmann> | where you saw the bunch of boxes |
| 00:59 | <erlehmann> | hyperlinks are “text as control” btw |
| 01:31 | <JonathanNeal> | erlehmann: if there is a way to do this in windows, mac, and linux, then awesome. |
| 01:32 | <erlehmann> | JonathanNeal if you want to have up to date unicode coverage, install unifont. |
| 01:32 | <erlehmann> | here http://unifoundry.com/unifont.html |
| 01:32 | <erlehmann> | :) |
| 01:32 | <erlehmann> | or better … ☺ |
| 01:32 | <JonathanNeal> | svgs are an excellent way of doing them as well, we can fallback on a 1x png |
| 01:33 | <JonathanNeal> | erlehmann: its unreasonable to load a font face just to get table sortables |
| 01:34 | <erlehmann> | JonathanNeal of course it is. but the pointing triangles (▲▼) are widely supported |
| 01:34 | <erlehmann> | they are in unicode since whatever |
| 01:34 | <erlehmann> | better than SVG |
| 01:34 | <erlehmann> | http://en.wikipedia.org/wiki/Geometric_Shapes |
| 01:35 | <erlehmann> | > ^ The Unicode Standard Version 1.0, Volume 1. Addison-Wesley Publishing Company, Inc. 1990, 1991. ISBN 0-201-56788-1. |
| 01:35 | <erlehmann> | JonathanNeal if 20+ years of unicode is not good enough, then i don't know! |
| 01:36 | <JonathanNeal> | got it |
| 01:39 | <erlehmann> | :) |
| 01:40 | <erlehmann> | hey, did you know you can make almost everything seem slightly passive-aggressive if you append a smilie? :) |
| 01:44 | <JonathanNeal> | erlehmann: sorry if i was communicating poorly, i was watching my kids between responses |
| 01:44 | <erlehmann> | JonathanNeal no problem. i am appreciating not having to write a polyfill myself! |
| 01:45 | <JonathanNeal> | erlehmann: i would love to see your other work. github? twitter? |
| 01:45 | <erlehmann> | i rarely use github since they blocked by account once |
| 01:45 | <erlehmann> | but here https://github.com/erlehmann |
| 01:46 | <erlehmann> | there is also http://news.dieweltistgarnichtso.net/ and http://daten.dieweltistgarnichtso.net/src/ |
| 01:46 | <erlehmann> | JonathanNeal the most useful thing i ever wrote was probably this https://github.com/wpoa/open-access-media-importer |
| 01:47 | <erlehmann> | it is a bot that uploads audio and video files to wikimedia commons |
| 01:47 | <erlehmann> | the most interesting thing i wrote was probably libglitch http://news.dieweltistgarnichtso.net/talks/music-c-compiler.html |
| 01:49 | <erlehmann> | if you want to make similarly useless data uris, i have a CLI alternative to Hixie's data uri kitchen! http://news.dieweltistgarnichtso.net/bin/file2datauri.html |
| 01:49 | <erlehmann> | i should really collect all that stuff |
| 01:49 | <erlehmann> | in one place |
| 01:50 | <erlehmann> | JonathanNeal do you have a portfolio type site? |
| 01:55 | <JonathanNeal> | github.com/jonathantneal |
| 01:56 | <erlehmann> | if you are into electronic music, watch my talk about making music with a c compiler! |
| 01:57 | <JonathanNeal> | and if you’re into code, listen to my electronic music about the internet http://jonneal.bandcamp.com/ |
| 01:58 | <erlehmann> | > Paul Irish's jQuery Anti-Pattern for Performance of the Week 00:07 |
| 01:58 | <JonathanNeal> | That will change your life. |
| 01:58 | <erlehmann> | i almost laughed :D |
| 01:58 | <JonathanNeal> | It is the soundtrack of a podcast http://yayquery.com/ |
| 01:59 | <erlehmann> | if you can speak german, i also did a podcast some time ago |
| 01:59 | <erlehmann> | http://warumnicht.dieweltistgarnichtso.net |
| 02:00 | <erlehmann> | a podcast about … jquery |
| 02:00 | <erlehmann> | how odd |
| 02:01 | <kriskowal_> | whynot.theworldisnotlikethatatall.net/itisalldifferent |
| 02:02 | <erlehmann> | hehe |
| 02:03 | <JonathanNeal> | Well, wife is back to watch the kids, so now I can get back to coding that polyfill. erlehmann, i always release cc0 so remix it as you like. |
| 02:03 | <erlehmann> | kriskowal_ here :D http://daten.dieweltistgarnichtso.net/tmp/etc_passwd |
| 02:04 | <erlehmann> | it gives |
| 02:04 | <erlehmann> | HTTP/1.1 302 Found |
| 02:04 | <erlehmann> | Location: file:///etc/passwd |
| 02:04 | <erlehmann> | if i remember correctly, python's urllib could be fooled that way |
| 06:11 | zcorpan | notices poster="" doesn't use crossorigin |
| 06:13 | <zcorpan> | i guess the poster frame is never painted on a canvas so it's just the dimensions that leak but you can get them with <img> already |
| 07:35 | <MikeSmith> | so Android apparently provides an API that's smart enough to detect if you're on a metered connection or not |
| 07:35 | <MikeSmith> | http://stackoverflow.com/questions/23877476/android-why-connectivitymanager-isactivenetworkmetered-always-returning-true/23879816#23879816 |
| 07:36 | <MikeSmith> | by way of Jonas |
| 07:36 | <MikeSmith> | marcosc: I guess you already know about that |
| 07:37 | <SamB> | hmm, I guess it's not *that* surprising that your phone would be aware of how it is billed |
| 07:37 | <MikeSmith> | (relevant to the network-information API work) |
| 07:38 | <MikeSmith> | SamB: there are lots of things your phone knows that are aren't exposed to web content |
| 07:38 | <SamB> | oh, is that one, then? |
| 07:38 | <MikeSmith> | it has been in the past |
| 07:38 | <foolip> | annevk: fullscreen this week? |
| 07:39 | <MikeSmith> | SamB: and still is have iOS afaik -- that is, there's no iOS native API that provides the same info at that Android API does. Or maybe there is and I just missed the news |
| 07:40 | <SamB> | I suppose most users *would* probably prefer the content be asked to avoid frivolous bandwidth usage |
| 07:40 | <SamB> | MikeSmith: what, sense? iphone? how are those related? |
| 07:41 | <MikeSmith> | don't understand what you're asking |
| 07:42 | SamB | is not an apple fanboy |
| 07:42 | <MikeSmith> | if the iphone SDK doesn't provide a native API that the browser could use, then the browser has nothing it could use to expose the info to Web content |
| 07:43 | <MikeSmith> | is what I meant |
| 07:43 | <SamB> | sure |
| 07:44 | <SamB> | though, I'm not sure how much of an issue that would actually be if Apple's browser people wanted to make it happen ... |
| 08:02 | <annevk> | foolip: think so, also need to look at notifications |
| 08:28 | <foolip> | annevk: cool! |
| 08:58 | <zcorpan> | Hixie: as for names, i thought dashes were avoided for new attributes |
| 09:20 | <mounir_> | annevk: can you have another look at https://github.com/w3c/screen-orientation/pull/58 |
| 09:33 | <annevk> | mounir_: it seems okayish |
| 09:33 | <annevk> | mounir_: it seems it would be better if we defined actual hooks |
| 09:35 | <mounir_> | annevk: if you want to do that, I wouldn't mind :) |
| 09:40 | <annevk> | mounir_: I recommend adding an XXX comment for now |
| 09:40 | <annevk> | mounir_: we can do it once things around animation frame tasks are a bit more settled |
| 09:41 | <annevk> | I guess I can add a note at some point in Fullscreen too so I don't forget |
| 09:49 | <mounir_> | annevk: I can open an issue marked as Future |
| 09:51 | <annevk> | k |
| 09:51 | <annevk> | zcorpan: poster not using crossorigin seems like an oversight |
| 09:52 | <mounir_> | annevk: https://github.com/w3c/screen-orientation/issues/62 |
| 09:52 | <zcorpan> | annevk: yeah likely. but not sure we should change it now |
| 09:53 | <annevk> | zcorpan: it would mean e.g. that if you have <video crossorigin> the poster cannot be retrieved over the same connection |
| 09:53 | <annevk> | zcorpan: because it would be retrieved with cookies |
| 09:53 | <zcorpan> | annevk: yeah |
| 09:53 | <annevk> | zcorpan: seems bad for perf |
| 09:54 | <zcorpan> | annevk: wait i don't follow why it can't use the same connection |
| 09:55 | <annevk> | zcorpan: https://www.w3.org/Bugs/Public/show_bug.cgi?id=26556 |
| 10:02 | <zcorpan> | annevk: is it like this: <video poster> makes a new fetch, with credentials (username/password?), then <video src crossorigin=anonymous> makes a new fetch that has to start a new connection to omit the credentials |
| 10:03 | <mathiasbynens> | annevk, for context: I’m working on this ES6 Unicode regular expression transpiler http://mothereff.in/regexpu (and I want it to be correct) |
| 10:04 | <annevk> | zcorpan: yeah, "omit credentials" is not fully defined at the moment, it actually ensures to not reuse a connection so the server is less clear on it being the same client |
| 10:05 | <zcorpan> | annevk: ok |
| 10:05 | <annevk> | mathiasbynens: in the meeting I was in, all extensions were out because nobody had time to work them out |
| 10:06 | <annevk> | mounir_: see also http://annevankesteren.nl/2014/02/monkey-patch |
| 10:08 | <zcorpan> | annevk: <video poster crossorigin=anonymous> is probably used pretty rarely so maybe we can still change it |
| 10:09 | <annevk> | zcorpan: it would affect any poster crossorigin combination |
| 10:09 | <annevk> | zcorpan: poster not having CORS would prevent it from being displayed |
| 10:11 | <zcorpan> | yeah true |
| 10:11 | <annevk> | I like all these rumors on reversible USB cables, hope they become mainstream soonish |
| 10:13 | <mounir_> | annevk: I'm fully aware of the monkey patch article |
| 10:16 | <annevk> | mounir_: the reason I'm pointing to it is that you have not filed bugs for hooks or requested specific hooks |
| 10:18 | <mounir_> | annevk: it did not feel that much needed for fullscreen |
| 10:18 | <mounir_> | I did for the HTML specification |
| 10:22 | <mathiasbynens> | annevk: did you ever see encodings like `utf-8859-1` in the wild? |
| 10:22 | <mathiasbynens> | at first glance, it looks like that should be `iso-8859-1` which is a label for `windows-1252` |
| 10:23 | <mathiasbynens> | https://github.com/ForbesLindesay/legacy-encoding/issues/1#issuecomment-53221336 |
| 10:25 | <mathiasbynens> | maybe those should be added as labels? |
| 10:36 | <annevk> | mathiasbynens: only if we can be sure that would not degrade user experience |
| 10:37 | mathiasbynens | files bug |
| 10:37 | <annevk> | mathiasbynens: sites are known to have labels such as euc_jp and if you treat that as euc-jp you display the site wrong |
| 10:39 | <mathiasbynens> | annevk: oh wow :( |
| 10:39 | <mathiasbynens> | anyway, filed https://www.w3.org/Bugs/Public/show_bug.cgi?id=26655#c0 |
| 10:40 | <annevk> | not sure what I can do with that bug |
| 10:40 | <zcorpan> | annevk: possibly browsers can't use the same connection for video anyway because video does things like abort the connection and start a new one when seeking (and OGG seeks to the end when loading to get the duration) |
| 10:47 | <foolip> | zcorpan: seems like an oversight that crossorigin doesn't apply to the poster, no? |
| 10:47 | <annevk> | zcorpan: not even with HTTP/2? |
| 10:47 | <zcorpan> | foolip: sure |
| 10:47 | <zcorpan> | annevk: dunno |
| 10:48 | <zcorpan> | on github i only find test cases and demos that use both crossorigin and poster |
| 10:51 | <foolip> | I'd be somewhat surprised if there's any content that depends on <video crossoriginal=""> at all |
| 10:51 | <foolip> | heh, -al |
| 10:56 | <zcorpan> | mathiasbynens: in webdevdata from last year i see two instances of utf-8859-1 (in http content-type) that appear would have no difference if the label was supported since the correct name is in a <meta> |
| 11:18 | <mathiasbynens> | zcorpan: the link ^ has several examples where `utf-8859-1` (and others) are used in both the XML prolog and in <meta>, and the `Content-Type` header does not specify any encoding |
| 11:21 | <zcorpan> | mathiasbynens: all www-style stuff? |
| 11:21 | <mathiasbynens> | yep |
| 11:22 | <mathiasbynens> | mails sent by different people, though |
| 11:23 | <zcorpan> | weird. anyway, would need research what the web compat situation is like outside www-style before changing the spec |
| 11:31 | <Manishearth> | Ms2ger: what do you think of https://blog.mozilla.org/privacy/2014/07/22/prefersafe-making-online-safety-simpler-in-firefox/ ? |
| 11:32 | <Manishearth> | Firstly, how often does Mozilla do the implement-first-standards-later thing? It's a bit annoying :s |
| 12:09 | <annevk> | Manishearth: seems there's a draft; like everyone else Mozilla implements plenty of those |
| 12:12 | <Manishearth> | annevk: ah, okay |
| 12:52 | <aleray> | hi, I'm working with not that consistent rdf data and a same resource might be expressed with a trailing slash or not, and with the http or https protocole. For instance "http://example.org/triples.rdf"; or "https://example.org/triples.rdf/"; are the same resource. How can I construct sparql requests that query all variants? |
| 12:52 | <aleray> | or is there a way to normalize uri with rdflib? |
| 12:53 | <Ms2ger> | I don't think you want to be asking RDF questions here |
| 12:53 | m4nu | grins. |
| 12:53 | <m4nu> | hey aleray, you should probably ask that question on #swig on irc.w3.org - let me get some more info for you. |
| 12:54 | <m4nu> | the short answer is, yes there are ways to construct SPARQL queries to match that sort of data. |
| 12:55 | <m4nu> | aleray: sorry, it's #swig (Semantic Web Interest Group) on irc.freenode.net |
| 12:55 | <m4nu> | They'll be able to help you out. |
| 12:57 | <aleray> | m4nu, Ms2ger thanks. Waiting for an answer there |
| 12:57 | <aleray> | on the rdflib level there is nothing to do to normalize at parsing time? |
| 12:57 | <aleray> | oups |
| 12:58 | <aleray> | sorry those questions where meant for #rdflib |
| 12:58 | <aleray> | just noticed |
| 13:43 | <smaug____> | annevk: what you mean with "as the events go to the service worker directly." |
| 14:09 | <annevk> | smaug____: see the proposal, the idea is for the service worker to support notification events |
| 14:17 | <smaug____> | annevk: proposal where? |
| 14:18 | <smaug____> | mailing list? |
| 14:21 | <smaug____> | ok, if that prevents several events for one user input, fine |
| 14:47 | <annevk> | smaug____: yeah, linked from the GitHub issue |
| 14:49 | <smaug____> | annevk: we may still need some kind of list of open Notifications |
| 14:49 | <smaug____> | which wouldn't clone |
| 14:49 | <smaug____> | I think that would let b2g to have the Notification open even while the app is restarted and then get access to the Notification |
| 14:50 | <smaug____> | or something close to that... |
| 14:50 | <smaug____> | hmm |
| 14:51 | <smaug____> | a bit odd too, since in some cases it would create new Notification objects... |
| 14:51 | <smaug____> | but maybe it isn't too bad |
| 15:01 | <smaug____> | hmm, no, that wouldn't work |
| 15:48 | <Hixie> | zcorpan: yeah, the names in that e-mail are mostly placeholders until someone can give better names |
| 18:08 | <jsbell> | Apropos of nothing: over the weekend, between dealing with grumpy children, I converted a hobby project's use of XHR over to a Fetch prollyfill. |
| 18:11 | <caitp> | neat |
| 18:14 | <jsbell> | Converting non-200s to rejections seemed to be a common pattern in adapting the code, since my old code wanted error callbacks. Also, network errors as TypeError (spec section 5.6) seems a bit odd, but it seems like we're using TypeError for everything these days. |
| 18:16 | <jsbell> | And on that note: I wonder if we can get away with converting some uses of DataError to TypeError in the IDB spec - e.g. for invalid keys - so we can replace some prose with better IDL. |
| 18:18 | <caitp> | > network errors as TypeError (spec section 5.6) seems a bit odd, but it seems like we're using TypeError for everything these days. |
| 18:19 | <caitp> | that does sound profoundly odd |
| 18:19 | <caitp> | assuming you mean network errors as in status === 0 |
| 18:19 | <caitp> | although it would be just as odd for status === 400 |
| 18:20 | <jsbell> | yes, for status === 0. Assuming I'm reading the spec right. |
| 18:22 | <Domenic> | The idea is fetch() is something that could potentially be useful in more than just web environments |
| 18:22 | <Domenic> | And ES only provides a few error types, with TypeError being the catch-all |
| 18:24 | <caitp> | there's no reason DOMExceptions / Errors couldn't be defined in ES, there's nothing really web-specific in them |
| 18:24 | <jsbell> | Yeah; I can usually squint and say "I guess that's sort-of a 'you passed the wrong sort of thing here', so TypeError makes sense". This one stood out, but *shrug*. |
| 18:25 | <jsbell> | Eh, I don't want ES littered with more new error types. |
| 18:51 | <gsnedders> | I'd like more error types, but hey. |
| 19:13 | <annevk> | jsbell: since we don't reveal types of network errors, TypeError seemed fine |
| 19:13 | <annevk> | jsbell: I guess the only downside would be that you also get that for incorrect usage of the API |
| 19:14 | <jsbell> | I don't have a better suggestion. |
| 19:23 | <Domenic> | JakeA et al: does Cache have any sense of URL normalization? E.g. storing a request with URL example.com/ vs. example.com ? |
| 19:26 | <JakeA> | Domenic: good point, we should make .add something sensible with redirects |
| 19:26 | <JakeA> | Although example.com will be the same as example.com/ |
| 19:27 | <Domenic> | right, i guess my quesiton is, to what extent is Cache just a key/value store, vs. a canonicalized-URL/value store. |
| 19:31 | <JakeA> | It's more than key-value. The matching is like the browser cache, but better (stores entries per vary variation) |
| 20:17 | <mathiasbynens> | redirects aside, i’d expect the same URL normalization that is being used everywhere else on the platform, i.e. `function normalize(url) { var el = document.createElement('a'); el.href = url; return el.href; }` |
| 20:27 | <zcorpan> | mathiasbynens: document encoding or utf-8 for the query? |
| 20:28 | <mathiasbynens> | zcorpan: doesn’t the above always use utf-8? |
| 20:29 | <zcorpan> | no |
| 20:46 | <mathiasbynens> | woah |
| 20:48 | <mathiasbynens> | zcorpan, do you happen to have a link/example? |
| 20:48 | <Hixie> | annevk: "flush out" means "empty", "flesh out" means "fill", fwiw. |
| 20:49 | <caitp> | short of the custom elements stuff, was there ever a proposed JS api for enqueueing microtasks? |
| 20:50 | <jsbell> | caitp: Not a serious one that I recall... |
| 20:51 | <caitp> | hm |
| 20:51 | <Hixie> | caitp: create a promise and resolve it straight away? |
| 20:51 | <erlehmann> | so this still works? http://gsnedders.html5.org/outliner/process.py?url=http%3A%2F%2Fdaten.dieweltistgarnichtso.net%2Ftmp%2Fetc_passwd |
| 20:51 | <erlehmann> | gsnedders helo |
| 20:58 | <Domenic> | yes, but ideally we should have global.asap(function () { }) instead of having to do Promise.resolve().then(function () { })... so silly. |
| 21:00 | <jgraham_> | erlehmann: What's the problem? |
| 21:01 | <erlehmann> | http://daten.dieweltistgarnichtso.net/tmp/dev_urandom |
| 21:01 | <erlehmann> | http://gsnedders.html5.org/outliner/process.py?url=http%3A%2F%2Fdaten.dieweltistgarnichtso.net%2Ftmp%2Fdev_urandom |
| 21:01 | <erlehmann> | jgraham_ gsnedders outliner accepts 302 redirects to file: URLs |
| 21:13 | <annevk> | Hixie: need some context |
| 21:13 | <annevk> | Hixie: https://twitter.com/brucel/status/503900281089032192 |
| 21:14 | <Hixie> | annevk: in some bug you had said we needed to "flush out" some step or other |
| 21:15 | <annevk> | ah, thanks, I should've known that :-) |
| 22:53 | <caitp-> | Hixie: it kind of amazes me that I'm saying this, but one of the nice things about the ES specs is the tendency to give blocks in instructions (like loops, or case statements, etc) into nested lists, so it's a bit easier to read and visualize. It would be really nice if that could be done for the html stuff ._. |
| 22:54 | <caitp-> | as it is I usually end up copy/pasting it into a text editor and laying it out like that manually, which is time consuming :> |
| 22:55 | <Hixie> | html does do that quite a lot |
| 22:55 | <Hixie> | if you have any specific cases that are hard to understand, file a bug |
| 22:55 | <caitp-> | will do |
| 23:07 | <gsnedders> | erlehmann: I know |
| 23:08 | <gsnedders> | erlehmann: it's not my server, and nobody's bullied me into fixing it yet, and I've been pretty busy :P |
| 23:08 | <gsnedders> | erlehmann: pull requests welcome! |
| 23:08 | <erlehmann> | gsnedders how could i exploit it? |
| 23:08 | <gsnedders> | erlehmann: mostly it seems like private html files on the server are the risk |
| 23:09 | <gsnedders> | erlehmann: not that likely to be many |
| 23:11 | <erlehmann> | hmm ._. |
| 23:11 | <gsnedders> | erlehmann: still absolutely should be fixed |
| 23:22 | <erlehmann> | gsnedders just upgrade python.j |
| 23:22 | <erlehmann> | i bet the new urllib has it fixed. |
| 23:33 | <gsnedders> | erlehmann: I don't want to have to maintain my own version of Python! |
