| 08:41 | <annevk> | Domenic: so even if you use for correctly you can still get clashes if two standards define the same term |
| 08:41 | <annevk> | Domenic: e.g., I found out CSS Display defines "element" in its introduction and then doesn't really use it... |
| 08:44 | <annevk> | CSS is such a mess: https://www.w3.org/TR/CSS/#css |
| 08:45 | <annevk> | tantek: when are you getting HTTPS? |
| 08:46 | <annevk> | tantek: apart from a namespace (which cannot be upgraded to HTTPS) you're the only non-HTTPS URL in Fullscreen |
| 09:53 | <annevk> | beverloo: I'm doing some cleanup of Notification's bikeshed usage |
| 09:53 | <annevk> | beverloo: after that I can probably remove the two features you mentioned |
| 09:54 | <annevk> | beverloo: I also plan to setup the snapshot tooling |
| 10:12 | <tobie> | annevk: do me a favor and please give me thumbs up or down for https://gist.github.com/tobie/870d8ee05e1b34f00558667d4587fa75 |
| 10:12 | <tobie> | annevk: text for the issue I'll be opening against the various UAs re the regex removal |
| 10:14 | <annevk> | tobie: looks good, but I'd include a link to an issue to coordinate on tests |
| 10:14 | <annevk> | tobie: or if tests are already updated, note that somehow |
| 10:14 | <tobie> | I couldn't find any tests about this. |
| 10:15 | <annevk> | tobie: then probably easiest to file an issue against web-platform-tests and note that in these bug reports |
| 10:15 | <tobie> | annevk: it seems like the issue should be named: "write tests for the WebIDL spec" |
| 10:15 | <annevk> | tobie: heh, we have some right? |
| 10:16 | <tobie> | annevk: like three and a half |
| 10:16 | <annevk> | http://w3c-test.org/WebIDL/ is not empty |
| 10:16 | <Ms2ger> | There's idlharness |
| 10:16 | <Ms2ger> | And then there's a bunch of crappy, semi-reviewed tests in a PR somewhere |
| 10:16 | <annevk> | Ms2ger: where is that hosted? |
| 10:17 | <annevk> | tobie: anyway, make note of that somehow |
| 10:17 | <Ms2ger> | http://w3c-test.org/dom/interfaces.html |
| 10:17 | <tobie> | Ms2ger: well--that's not really a test for WebIDL or its bindings |
| 10:17 | <Ms2ger> | Sure it is |
| 10:17 | <annevk> | Yeah it is actually |
| 10:17 | <Ms2ger> | WebIDL only exists in so far as it's used in particular interfaces |
| 10:17 | <annevk> | That would be great to extend with tests for RegExp |
| 10:18 | <Ms2ger> | Unfortunately it's pretty hard to write tests for "this does not throw" in that generality, because the interfaces themselves can throw |
| 10:19 | <annevk> | hmm fair |
| 10:20 | <tobie> | well, it's indirect and makes a large number of assumptions. But I guess that's the best we can do? |
| 10:22 | <annevk> | You can't really test a framework directly |
| 10:39 | <annevk> | beverloo: https://github.com/whatwg/notifications/pull/85 |
| 12:20 | <nox> | :scope is dead, right? |
| 12:43 | <annevk> | It is? |
| 13:06 | <nox> | annevk: Brainfarted. |
| 13:51 | <Domenic> | annevk: that's what link-defaults is supposed to be able to fix |
| 13:51 | <Domenic> | annevk: but you need the new Assume Explicit For: yes for it to really work |
| 13:51 | <Domenic> | annevk: but that has some bugs I think, still shaking them out |
| 14:13 | <annevk> | Domenic: I think SW is using Bikeshed way wrong |
| 14:14 | <annevk> | Domenic: but I guess you can't fix it all in one go |
| 14:14 | <Domenic> | annevk: yeah, exactly. I think part of it is that people thought they needed a giant anchors block for everything because you needed one for HTML |
| 14:14 | <annevk> | Domenic: Notifications API was quite the mess too, just put a lot effort into it earlier today |
| 14:14 | <Domenic> | Now that we've fixed that to be mostly not true for HTML we'll be in a better place |
| 14:14 | <Domenic> | Yeah I saw that; quite nice |
| 14:15 | <annevk> | I do think we're missing a lot by not having output diffs in the new world |
| 14:15 | <annevk> | I realize a lot of standards are written without them, but the output diff shows you forgotten end tags, links going to the wrong place, etc. |
| 14:16 | <Domenic> | So it's pretty doable to get |
| 14:16 | <Domenic> | Let me try something |
| 14:17 | <Domenic> | So here is the output generated from the bs file: https://api.csswg.org/bikeshed/?url=https://raw.githubusercontent.com/whatwg/dom/89a849a54f123b3409be4bf371735fd0f901ef20/dom.bs |
| 14:18 | <Domenic> | Here's a HTML diff: http://services.w3.org/htmldiff?doc1=https%3A%2F%2Fapi.csswg.org%2Fbikeshed%2F%3Furl%3Dhttps%3A%2F%2Fraw.githubusercontent.com%2Fwhatwg%2Fdom%2F89a849a54f123b3409be4bf371735fd0f901ef20%2Fdom.bs&doc2=https%3A%2F%2Fdom.spec.whatwg.org%2F |
| 14:18 | <Domenic> | If we could find a simple text differ service we'd also be done |
| 14:19 | <annevk> | Domenic: would that help if just the href value changed? |
| 14:19 | <Domenic> | annevk: I don't know if HTMLdiff handles that, but a text differ would |
| 14:19 | <annevk> | Domenic: yeah, I think I'm mostly interested in a `diff -up` |
| 14:20 | <Domenic> | Something like https://www.diffnow.com/?report=zs9vq but not as limited |
| 14:21 | <Domenic> | We could have a bot like tobie has do the work and upload it to a GitHub gist |
| 14:37 | <tobie> | I could make the existing bot run for a number of repos. |
| 14:37 | <tobie> | shouldn't be too difficult. |
| 14:38 | <Ms2ger> | annevk, aloha |
| 14:38 | <annevk> | Ms2ger: que? |
| 14:40 | <Ms2ger> | annevk, so Servo has a bunch of custom about: URLs, and I'm wondering where best to implement them... Fetch has this note "URLs such as "about:config" are handled during navigation and result in a network error in the context of fetching.", but I can't find where in the navigation algorithm that is |
| 14:44 | <annevk> | Ms2ger: wherever custom schemes are handled presumably, but there might be a bug |
| 16:05 | <tantek> | annevk: good q. have been trying to get my hosting provider to provide a LetsEncrypt option but I may just end up doing it manually |
| 18:14 | <TabAtkins> | Domenic: Per #862, it's just markup errors causing you a problem. |
| 18:24 | <Domenic> | TabAtkins: thanks! |
| 18:54 | <smaug> | is it defined somewhere how contentEditable and shadow DOM should work together? |
| 20:18 | <noah> | looking at https://github.com/w3c/web-platform-tests/blob/e7da296741/url/urltestdata.json#L2147 ... why is it that "hash" is "" even though href ends with "#"? whenever there's a non-empty fragment, "#" is included in the value of "hash" |
| 20:20 | <rbyers> | annevk, Domenic: you guys want to encourage participation in the HTML spec by web developers, right? |
| 20:20 | <rbyers> | .. Eg. if at CDS I say we're getting better at involving web developers in the API design / validation process, it's cool to include a screenshot of the html github repo page or something? |
| 20:21 | <rbyers> | .. or maybe https://github.com/whatwg/html/blob/master/CONTRIBUTING.md ? |
| 20:22 | <Domenic> | rbyers: yes! :D |
| 20:24 | <rbyers> | Ok, I'll plan on it. |
| 20:27 | <Domenic> | rbyers: if it helps, we currently have 78 distinct commiters, compared to 1 a little over a year ago |
| 20:27 | <rbyers> | Ooh that's a fun stat |
| 20:27 | <rbyers> | Thanks |
| 20:28 | <rbyers> | Was it really true that Hixie was the only person that modified HTML up until a year ago? |
| 20:28 | <Domenic> | According to commit history yes, although it's possible other people submitted text or even HTML which he incorporated |
| 20:29 | <rbyers> | ok |
| 20:29 | <Domenic> | Hmm wait I think it was 2, zcorpan's <picture> stuff was being pulled in via a build script |
| 20:29 | <rbyers> | hah ok, still a good stat. I'll see if I can fit it in. |
| 20:30 | <rbyers> | General theme of this part of the talk: |
| 20:30 | <rbyers> | "The web platform is unique because it's a commons - we all own it collectively, and it thrives on collaboration and a healthy marketplace of competing ideas. So it's very important that there's great communication between everyone in the ecosystem." |
| 20:32 | <rbyers> | Folks here would probably also love the quote I just read: https://twitter.com/RickByers/status/794517804300914688 |
| 20:32 | <rbyers> | From one of the TCP/IP Engineers commenting on TCP/IP over ISO stack: "Standards should be discovered, not decreed. Seldom has it worked any other way." |
| 20:35 | <Domenic> | Yeah I saw that, pretty great :D |
| 20:35 | <Domenic> | Sounds like a fun talk! |
| 20:41 | <annevk> | noah: silly legacy |
| 20:42 | <annevk> | noah: a lot of APIs are somewhat poorly designed in hindsight |
| 20:43 | <annevk> | rbyers: sounds great |
| 20:43 | <noah_> | annevk: ok |
| 20:43 | <annevk> | rbyers: note that for TCP/IP I heard you cannot implement that from standards as they don't reflect reality 😟 |
| 20:44 | <rbyers> | Hah, maybe that's a problem with "discovering" a standard - you don't always "discover it" correctly... |
| 20:44 | <annevk> | noah_: this is where the topic originates |
| 20:45 | <annevk> | rbyers: yeah, guess they missed the iteration aspect |
| 21:17 | <noah> | annevk: sorry i keep getting disconnected, i think i missed what you said was the origin of the thing |
| 21:21 | <noah> | somewhere in https://url.spec.whatwg.org/ there must be a step that replaces the empty path with "/", e.g. "http://example.com"; => "http://example.com/";, but i'm having trouble locating it, anybody know? |
| 21:33 | <annevk> | noah: serializer? |
| 21:35 | <noah> | annevk: ahh yeah |
| 23:36 | <nox> | https://twitter.com/nokusu/status/794684470586707968 |
| 23:42 | <TabAtkins> | Uh, no, it's California Style Sheets. |