| 01:36 | <MikeSmith> | rniwa: do you need anything in https://github.com/w3c/web-platform-tests/pull/4016 specfically (re)reviewed? |
| 01:36 | <rniwa> | MikeSmith: not really. if you see any typos, etc... I can fix them |
| 01:36 | <rniwa> | MikeSmith: but the tests have been reviewed when it got landed into WebKit repo |
| 01:37 | <rniwa> | MikeSmith: if you're reviewing that one, I also have https://github.com/w3c/web-platform-tests/pull/3998 and https://github.com/w3c/web-platform-tests/pull/3968 |
| 01:40 | <MikeSmith> | rniwa: well I was asking just because our policy for wpt is that if tests have been reviewed by a browser project upstream, then we can just merge them |
| 01:40 | <MikeSmith> | so I will just merge #4016 now, and look at the others |
| 01:41 | <rniwa> | MikeSmith: thanks |
| 01:41 | <rniwa> | MikeSmith : https://github.com/w3c/web-platform-tests/pull/3968 is the only one that hasn't been reviewed in WebKit |
| 01:41 | <MikeSmith> | OK |
| 01:43 | <MikeSmith> | for https://github.com/w3c/web-platform-tests/pull/3998 I guess before merging it we should wait for Domenic to LGTM the changes you made in response to his review comments |
| 03:57 | <peaches> | hello |
| 05:32 | <annevk> | peaches: morning |
| 05:54 | <rniwa> | MikeSmith: maybe you can review https://github.com/w3c/web-platform-tests/pull/4019 ?\ |
| 05:54 | <rniwa> | annevk: hi annevk! |
| 05:55 | <rniwa> | annevk: it's a brand new Wednesday, and we've got a brand new custom elements issue for you ;) https://github.com/w3c/webcomponents/issues/590 |
| 06:00 | <annevk> | rniwa: hmm, I wonder how we can fix that |
| 06:00 | <rniwa> | annevk: I think we need to add another custom element state like "upgrading" |
| 06:00 | <annevk> | rniwa: ah, more state |
| 06:00 | <rniwa> | annevk: what's weird here is that |
| 06:00 | <rniwa> | annevk: if the upgrade then fails, we'd have to clear any such reaction that has been enqueued for that element |
| 06:01 | rniwa | adds more test cases to his todo list |
| 06:02 | <rniwa> | I do recall Domenic talking about having to clear the queue so I bet he thought this already works... |
| 06:05 | <rniwa> | annevk: oh actually, I might be reading this incorreclty |
| 06:05 | <rniwa> | annevk: i think we parse through attributes when we upgrade elements |
| 06:08 | <annevk> | rniwa: yeah, seems like this already works |
| 06:08 | <annevk> | rniwa: it might be worth adding a note to the clone algorithm though |
| 06:08 | <rniwa> | lol at <body/> |
| 09:06 | <annevk> | zcorpan: did you have a way to test "salvageable"? |
| 09:06 | <annevk> | zcorpan: I'd like to figure out https://github.com/whatwg/html/issues/1900 |
| 09:06 | <zcorpan> | annevk: there's something in wpt, hold on |
| 09:09 | <zcorpan> | annevk: i think http://web-platform.test:8000/websockets/unload-a-document/001.html |
| 09:16 | <annevk> | zcorpan: ta, that looks fairly complicated |
| 09:17 | <zcorpan> | yeah :-( |
| 09:21 | <zcorpan> | effectively if the document was discarded, the script runs again. so the test uses sessionStorage to check if it's run the first time or second time. and a setTimeout for the case where navigating back to a non-discarded document |
| 09:22 | <zcorpan> | and it needs to do that in a popup window since testharness.js can't have the test itself be navigated |
| 09:26 | <annevk> | thanks, I guess I can use that setup |
| 09:26 | <annevk> | zcorpan: so all browsers implement that concept? |
| 09:26 | <annevk> | zcorpan: seems to pass in Firefox and Chrome |
| 09:28 | <zcorpan> | annevk: this particular test would pass if a browser always discards on navigation. i'm not sure what the current status is |
| 09:29 | <zcorpan> | annevk: it seems http://web-platform.test:8000/websockets/unload-a-document/002.html fails in chrome (test expects not-discarded) |
| 09:30 | <zcorpan> | but also fails in firefox :-| |
| 09:34 | <annevk> | zcorpan: hah, guess we should revisit that then |
| 09:34 | <zcorpan> | yeah |
| 09:42 | <zcorpan> | filed an issue |
| 11:38 | <smaug> | zcorpan: curious, do you know if other browsers will align with innerText spec? |
| 12:09 | <zcorpan> | smaug: https://bugs.chromium.org/p/chromium/issues/detail?id=651764 https://bugs.chromium.org/p/chromium/issues/detail?id=573309 |
| 12:14 | <zcorpan> | smaug: https://bugs.webkit.org/show_bug.cgi?id=160971 |
| 12:14 | <smaug> | k |
| 12:17 | <zcorpan> | i suppose some things are still subject to change in the spec, like throwing for void elements, since the majority do that |
| 12:58 | <annevk> | zcorpan: ping https://github.com/whatwg/url/issues/95 |
| 13:05 | <zcorpan> | annevk: commented |
| 13:05 | <zcorpan> | annevk: maybe i can PR that |
| 13:47 | <zcorpan> | annevk: hmmm. if the document encoding is utf-16, the spec says to percent-escape the query using utf-16, correct? afaict browsers percent-encode as utf-8 in that case |
| 13:48 | <zcorpan> | data:text/html;charset=utf-16,<%00!%00D%00O%00C%00T%00Y%00P%00E%00%%002%000%00h%00t%00m%00l%00>%00<%00a%00%20%00h%00r%00e%00f%00%3D%00"%00h%00t%00t%00p%00%3A%00%2F%00%2F%00e%00x%00a%00m%00p%00l%00e%00.%00o%00r%00g%00%2F%00&%00a%00u%00m%00l%00;%00%3F%00&%00a%00u%00m%00l%00;%00%23%00&%00a%00u%00m%00l%00;%00"%00>%00t%00e%00s%00t%00 |
| 13:48 | <annevk> | zcorpan: see get the output encoding |
| 13:48 | <Ms2ger> | Ah, the curse that keeps on giving |
| 13:48 | <zcorpan> | aha |
| 15:03 | <annevk> | Domenic: I guess we should figure out https://github.com/whatwg/html/issues/1698 first then |
| 15:03 | <annevk> | Domenic: I asked foolip if he could maybe do some code spelunking and explain what happens |
| 15:04 | <annevk> | Domenic: if we could remove that it would drastically simplify the model |
| 15:06 | <Domenic> | annevk: I mean I think that issue is just another Mozilla-vs.-the-world issue |
| 15:06 | <Domenic> | Although I haven't tested Edge |
| 15:07 | <annevk> | Domenic: I suspect so, I mostly want to know what happens exactly in the world so it can be defined (or explained why the world is wrong somehow) |
| 15:08 | <Domenic> | by "the world" you mean most browsers? Yeah. |
| 15:08 | <Domenic> | It would be much nicer if Edge were open source -_- |
| 15:08 | <foolip> | I'd like to have their IDL files :) |
| 15:08 | <annevk> | Domenic: I was using the world in the same manner as you :p |
| 15:08 | <Domenic> | oh right lol |
| 15:09 | <Domenic> | foolip: I feel like that is close... they expose a bunch of stuff on their website which is derived from the IDL. A short hop to the IDL itself. |
| 15:09 | <Domenic> | I guess probably more legal hurdles since it's part of Edge. |
| 15:10 | <foolip> | yeah, extended attributes and comments in there |
| 15:17 | <Ms2ger> | foolip, for https://github.com/w3c/web-platform-tests/pull/4001 , is that waiting on the spec PR? |
| 15:18 | <foolip> | Ms2ger: I was hoping for signoff from cdumez, see https://github.com/whatwg/html/issues/1519 |
| 15:19 | <foolip> | Ms2ger: do you want the change? |
| 15:20 | <Ms2ger> | foolip, just fighting my way through my inbox, and wasn't sure on the status |
| 15:21 | <foolip> | fighting through inbox, sounds vaguely familiar |
| 15:39 | <Ms2ger> | Does anyone besides Hixie_ have access to acid3? |
| 15:45 | <annevk> | Ms2ger: doubt it |
| 16:28 | <smaug> | foolip: outerText alias to innerText o_O |
| 16:28 | <smaug> | (getter) |
| 16:28 | <zcorpan> | smaug: yep. how would it be different? |
| 16:29 | <smaug> | what is the use for outerText? |
| 16:30 | <zcorpan> | for setting, replacing an element with some text. e.g. replacing images with their alt text maybe? |
| 16:30 | <smaug> | foolip: I would have imagined getter too would behave different related to newlines |
| 16:30 | <smaug> | differently |
| 16:31 | <smaug> | but ok, odd API, but should be too difficult to implement |
| 16:31 | <smaug> | assuming setter isn't too complicated |
| 17:30 | <younder> | This lack of MathML in blink is causing me greef. MathML is incomplete in WebKit. Nonexistant in blink. Complete in Firefox, but that has a crap unsupported embedded UI. |
| 18:28 | <Domenic> | annevk: creating "topic: events". We have a lot. |
| 18:31 | <annevk> | Domenic: cool |
| 18:46 | <Domenic> | WTF does step 11.1 mean in https://html.spec.whatwg.org/multipage/comms.html#dom-messageport-postmessage |
| 18:46 | <Domenic> | has been there forever https://html.spec.whatwg.org/commit-snapshots/c9e804f04d03a0658bfa689cb0f368a4d2e37936#dom-messageport-postmessage |
| 18:50 | <annevk> | Domenic: that is the lack of serialization and model to communicate across event loops |
| 18:51 | <Domenic> | How can events find themselves in task sources |
| 18:51 | <Domenic> | Especially since nobody ever puts them there |
| 18:52 | <annevk> | Domenic: the language is bogus too but you can't really fix it without serialization being a thing |
| 18:53 | <annevk> | Domenic: hard to explain how the structured cloned object gets created in the correct global too with this language |
| 18:54 | <annevk> | Domenic: so I just left it alone so far until we tackle the bigger problem |
| 18:54 | <annevk> | Technical debt 🎉 |
| 19:11 | <younder> | What are you numbasses going to do about MathML? (chromium, safari) |
| 19:11 | <younder> | nothing.. |
| 19:14 | <younder> | Well, I need it. looks like I have to write it myself. |
| 19:15 | <younder> | pointers (at least)? |
| 19:33 | <Domenic> | What am I missing about BroadcastChannel? http://jsbin.com/ruguhuwika/edit?html,console,output |
| 19:35 | <jyasskin> | younder: Insults aren't going to be very effective. But then, nothing individual is likely to be effective at getting MathML implemented. Try https://www.mathjax.org/. |
| 19:37 | <jyasskin> | Mek_: Are you the expert on BroadcastChannel? |
| 19:41 | <younder> | jyasskin, I am going to implement mathML. I'm not a whiner. I am a member of AMS (American Mathematical Society). I will submitt a branch, and then you can take it or leave ir. |
| 19:45 | <jyasskin> | younder: See https://groups.google.com/a/chromium.org/d/topic/chromium-dev/mSQVTZpdQc0/discussion for Chrome. You really would be better off using MathJax. |
| 19:46 | <younder> | jyasskin, it just isn't fast enough |
| 19:47 | <Domenic> | Sounds like a good open source project then to make Mathjax faster |
| 19:48 | <younder> | MathJax is javeScript. I need math in a editor powered by javascript but fast enough to allow interactive editing |
| 19:48 | <jyasskin> | younder: e.g. Maybe WebAssembly would speed it up enough. |
| 19:50 | <younder> | MathML is not that difficult. I'll just implement it. |
| 19:50 | <jyasskin> | Good luck. |
| 19:50 | <younder> | thx |
| 19:57 | <younder> | My platform for now is the blink engine. |
| 19:59 | <jyasskin> | You'll need to ask on blink-dev or chromium-dev for advice on that. I wouldn't expect your patch to get accepted into the main tree, but they might be able to point you toward the right way to start in a private fork. |
| 20:00 | <younder> | private fork it is. |
| 20:02 | <younder> | I was expecting that. I will only supply a port when it complies with al tests and is effectively merged with the latest version. |
| 20:03 | <younder> | al => all |
| 20:27 | <TabAtkins> | younder: You might be interested in KaTeX <https://khan.github.io/KaTeX/>;, a JS library from Khan Academy that does math rendering much faster than MathJax. |
| 20:27 | <TabAtkins> | (Partly because it purposely doesn't implement all of MathJax's functionality, but it's more than enough for most math.) |
| 20:34 | <younder> | TabAtkins, So like WebKit,I do string theory. It takes it to a different level. I also like category theory. |
| 20:35 | <younder> | I would like category diagrams. |
| 20:36 | <younder> | I'll figure it out. |
| 20:36 | <TabAtkins> | I would be rather surprised if MathML supports category diagrams? I'd look for a more dedicated renderer for those things. |
| 20:36 | <younder> | Just thought you would like a head start. |
| 20:37 | <younder> | TabAtkins, It doesn't |
| 20:38 | <jyasskin> | TabAtkins: I've got a patch marking certain sections "unstable" at https://github.com/WebBluetoothCG/web-bluetooth/pull/308. Do you like the rendering? Think it would look better some other way? Want it in bikeshed? |
| 20:39 | <TabAtkins> | jyasskin: Feels really distracting to me - I'm having trouble reading the algorithms. |
| 20:40 | <jyasskin> | TabAtkins: One problem I've found is that I can't use this technique to mark unstable bits of IDL. I'd need an extended attribute for that, but it's one that would never wind up in implementations. |
| 20:40 | <TabAtkins> | Versus effects like https://drafts.fxtf.org/paint/, where the color doesn't distract me. |
| 20:40 | <jyasskin> | Hm, ok. I wonder if it's the size, or if gray is just the wrong color. |
| 20:41 | <TabAtkins> | I think it's the size - too close to the size of normal text, and skinny serifs, so there's more intersections with the text. Huge chunky text is easier to filter out as a background detail. |
| 20:42 | TabAtkins | considers porting KaTeX to Python for Bikeshed... |
| 20:42 | <TabAtkins> | Unexpectedly, it can render headless without using any font information directly, so I really could just output its stuff from Python. |
| 20:43 | <TabAtkins> | jyasskin: Have you thought about a side-strip instead of a background, so it's still visible even when elements with backgrounds show up? |
| 20:43 | <TabAtkins> | Also, unknown extended attributes are accepted as valid and just left unparsed by widlparser, I believe. |
| 20:45 | <jyasskin> | TabAtkins: I'm not sure how to get a side strip to appear at a consistent place on the page regardless of the element I stick the class on. |
| 20:47 | <TabAtkins> | Ah, yeah, that's hard. |
| 20:52 | <jyasskin> | TabAtkins: Check https://api.csswg.org/bikeshed/?url=https://raw.githubusercontent.com/jyasskin/web-bluetooth-1/at-risk/index.bs#permission-api-integration again. |
| 20:53 | <TabAtkins> | I might go a little chunkier still, but *much* better. |
| 20:53 | <jyasskin> | I'm thinking of fixing the problem with notes covering the background, by using opacity for their backgrounds, instead of solid light colors. |
| 20:54 | <jyasskin> | There is no chunkier unless I pick a particular font. |
| 20:54 | <TabAtkins> | Is that not an image? |
| 20:54 | <jyasskin> | Nope, it's an inline SVG. |
| 20:54 | <TabAtkins> | Ah, yes, I see. |
| 20:55 | <jyasskin> | Amazingly, FF, Edge, and Chrome all render it without escaping. :) |
| 20:56 | <jyasskin> | (Unless I've broken something since I checked) |
| 20:56 | <TabAtkins> | Nothing too amazing; CSS parsing isn't that hard. ^_^ |