| 01:43 | <GPHemsley> | What's the current best practice for writing CSS tests? |
| 02:03 | <astearns> | GPHemsley: reftests for anything testing display results. testharness for parsing |
| 02:03 | <GPHemsley> | URL(s)? |
| 02:04 | <astearns> | http://testthewebforward.org/docs/reftests.html |
| 02:04 | <astearns> | http://testthewebforward.org/docs/testharness-tutorial.html |
| 02:04 | <astearns> | and please send feedback for improving those pages |
| 03:26 | <MikeSmith> | is anybody familiar with the background on http://web-platform.test/url/urltestdata.txt ? |
| 03:27 | <MikeSmith> | oops make that https://github.com/w3c/web-platform-tests/blob/master/url/urltestdata.txt |
| 03:28 | <SamB> | hmm, didn't someone recently say it came from webkit? |
| 03:29 | <MikeSmith> | yeah jgraham did |
| 03:30 | <MikeSmith> | anyway it seems to be assuming a base URL of http://example.org/foo/bar |
| 03:31 | <MikeSmith> | which testing it a more just now, I guess it is, consistently |
| 03:31 | <MikeSmith> | I had thought in some cases it was not but I guess I was mistaken |
| 03:47 | <SamB> | well, it looks like I was right and the WHATWG logo itself is not copyrightable: https://commons.wikimedia.org/wiki/Commons:Deletion_requests/File:WHATWG_logo_(Matthew_Raymond).png |
| 08:43 | <Ms2ger> | https://pbs.twimg.com/media/BiflGFdCcAERBtM.png:large |
| 08:46 | <ondras> | hurts. |
| 12:53 | <GPHemsley> | astearns (or anyone else): What's the policy on using examples from the spec as tests? |
| 12:54 | <Ms2ger> | If the spec isn't under some stupid copyright, it's clearly allowed |
| 12:57 | <GPHemsley> | "Copyright © 2014 W3C® (MIT, ERCIM, Keio, Beihang), All Rights Reserved. W3C liability, trademark and document use rules apply. " |
| 12:57 | GPHemsley | is not clear where that falls |
| 12:58 | <jgraham> | Examples from the spec probably aren't good tests |
| 12:59 | <jgraham> | Or at least they are such small subsets of good testsuites that just rewriting something using the same feature for the case that the example covers is easier than worrying about it |
| 13:03 | <Ms2ger> | GPHemsley, "dumb" |
| 13:04 | <GPHemsley> | :) |
| 13:16 | <smaug____> | any comments on having deltaX/Y and/or offset in scroll events? |
| 14:14 | <darobin> | GPHemsley: as indicated, examples are unlikely to be really good tests but they clearly fall under fair use so you're safe |
| 14:15 | <darobin> | technically, I think that as software they would be under the software license anyway, which is OSI |
| 14:28 | <gsnedders> | darobin: fall under fair use in some juristictions |
| 14:29 | <gsnedders> | (there exist juristictions with no concept of fair-use) |
| 14:29 | <darobin> | gsnedders: the W3C is judge of whether it's a problem or not, and it's clearly fair use under that jurisdiction :) |
| 14:30 | <gsnedders> | darobin: Assuming in all juristictions it is the W3C that has to file suit (I dunno if that's the case). |
| 14:30 | <darobin> | gsnedders: I don't know of a single jurisdiction in which IP issues are brought forth by the state rather than the wronged party |
| 14:31 | <darobin> | but I'm no expert |
| 14:31 | <darobin> | it seems unlikely |
| 14:31 | <darobin> | (even considering how fucked up IP legislation generally is) |
| 14:31 | <gsnedders> | darobin: copyright in the UK is a criminal matter, and certainly under English and Scottish law criminal cases aren't brought by the wronged party. |
| 14:32 | <darobin> | of course, you can claim (justly) that there is no guarantee that W3C will retain a sane policy forever — which is why the push for open licensing needs to continue |
| 14:32 | <darobin> | gsnedders: aren't there degrees? I forget, there's a name for the doctrine deciding how cases are brought about by the state |
| 14:33 | <darobin> | that said, I'm fairly sure that code examples actually fall under the software license |
| 14:34 | <gsnedders> | darobin: All criminal cases are brought to court by the CPS in English, and Procurator Fiscal in Scotland. I don't have the time to look into what copyright cases are criminal and what are not. |
| 14:45 | <gsnedders> | Okay, copyright is essentially just counterfeit goods and piracy. |
| 14:46 | <gsnedders> | (in the criminal sense) |
| 15:06 | <jgraham> | https://www.usenix.org/system/files/1403_02-08_mickens.pdf |
| 15:09 | <MikeSmith> | yeah that last third of so of that it pretty good |
| 15:09 | <MikeSmith> | he should have cut about half the jokes in their and it would have been a lot funnier |
| 15:10 | <MikeSmith> | maybe he got paid by the word |
| 15:29 | <gsnedders> | In the web case, there's nothing gained by impersonating a client in a TLS connection, as the client isn't authenticated, right? |
| 17:31 | <Hixie> | darobin: by and large, teh jurisdictions that have fair use are the same jurisdictions where fair use includes things required for interop, which would make all specs non-copyrightable :-P |
| 17:32 | <Hixie> | but don't tell jeff or rigo |
| 17:32 | <darobin> | Hixie: that is also my reading :) |
| 17:33 | <jgraham> | Presumably at most the normative sections |
| 17:34 | <jgraham> | So it would exclude examples |
| 17:35 | <darobin> | jgraham: I can think of quite a few specs in which the examples make the only part that enables interop :) |
| 17:37 | <jgraham> | "Your honor, this spec is a stinking pile of horse excrement and as-such doesn't actually allow interop without the examples" are words I want to hear in a courtroom drama |
| 17:47 | <JonathanNeal> | How do we markup svgs for screen readers? |
| 17:48 | <TabAtkins> | smaug____: +1 to deltaX/Y on scroll events. (Rough blink position, I think.) |
| 17:49 | <JonathanNeal> | SteveF: http://blog.paciellogroup.com/2011/08/html5-accessibility-chops-interactive-image-example/ any developments since this? |
| 17:50 | <JonathanNeal> | Is there a whatwg spec for this? I see http://www.w3.org/TR/SVG-access/ |
| 17:50 | <Hixie> | no |
| 17:51 | <JonathanNeal> | Bummer. Sorry for going off-topic then. |
| 17:56 | <jcgregorio> | Hixie: FYI https://codereview.chromium.org/178673002/ renames Path -> Path2D in canvas |
| 17:56 | <Hixie> | cool |
| 17:56 | <jcgregorio> | still hidden behind a flag though |
| 17:56 | <Hixie> | does it also make Path2D actually do what the spec says though? :-) |
| 17:56 | <Hixie> | unlike Safari's Path? :-) |
| 17:56 | <jcgregorio> | yes, it works like the spec says :-) |
| 18:00 | <jcgregorio> | doesn't implement addPathByStrokingPath or any of the add*Text methods |
| 18:02 | <Hixie> | cabanier: did you paste the right URL? I don't see anything from roc on that thread. |
| 18:02 | <Hixie> | jcgregorio: cool |
| 18:03 | <Hixie> | esprehn: btw the <dialog> focus stuff has landed in the spec. (not commenting on the blink-dev thread because every time i comment on it i seem to derail the intent-to-ship!) |
| 18:04 | <esprehn> | Hixie: I'll have to read it over, in the short term I'd rather we ship what we have and deal with the new stuff later |
| 18:04 | <Hixie> | yeah totally |
| 18:04 | <esprehn> | :) |
| 18:05 | <cabanier> | Hixie: sorry. copied the wrong line: https://groups.google.com/forum/#!topic/mozilla.dev.platform/5i8H-xKGtEg |
| 18:06 | <cabanier> | Hixie: I should update the wk implementation for Path |
| 18:06 | <cabanier> | Hixie: my patch for mozilla should also match the spec (except for not adding any of the add* methods) |
| 18:08 | <cabanier> | emails crossed :-) |
| 18:09 | <Hixie> | btw do you know what's the status of all the various canvas worker proposals? |
| 18:09 | <Hixie> | what's in the spec clearly isn't what people want, but i don't want to remove it then add something else later, that's twice the work of just adjusting it to be what people implement |
| 18:15 | <cabanier> | Hixie: No, I don't. Last I saw was roc's proposal and it sounded like someone was going to prototype it |
| 18:16 | <cabanier> | Hixie: but I didn't see anything land. I think it's fine to leave it in |
| 18:21 | <Hixie> | bummer |
| 18:22 | <Hixie> | we really need canvas in worker |
| 18:22 | <Hixie> | and i was really hoping someone would take care of it for me :-) |
| 18:24 | <cabanier> | yeah |
| 18:31 | <SamB> | Hixie: so what are you going to do, surround it with yellow tape and "dangerous bend" signs? |
| 18:32 | <Hixie> | which? the canvas in worker thing? |
| 18:32 | <SamB> | yeah that |
| 18:34 | <Hixie> | heh |
| 18:34 | <Hixie> | i dunno what to do |
| 18:35 | <Hixie> | what i'd like is to find something the vendors want to implement |
| 18:35 | <Hixie> | zcorpan: ping https://www.w3.org/Bugs/Public/show_bug.cgi?id=24835 |
| 18:53 | SamB | wonders if wikimedia commons will delete logo-xhr.svg if he uploads it ... |
| 18:54 | <SteveF> | JonathanNeal: http://blog.paciellogroup.com/2013/12/using-aria-enhance-svg-accessibility/ |
| 18:55 | <SteveF> | JonathanNeal:bottom line SVG accessibility will continue to be flaky until SVG 2 is implemented |
| 19:57 | <SamB> | you know what would be cool? if link targets would get highlighted when you just hover the link, not only after you follow them ... |
| 19:58 | <SamB> | would save time for those links that only go down a line or a paragraph or so |
| 20:05 | <Hixie> | interesting idea |
| 20:05 | <Hixie> | wonder how to do it |
| 20:06 | <SamB> | surely the CSS WG can figure out a way |
| 20:08 | <Ms2ger> | Ha |
| 20:10 | <JonathanNeal> | SteveF: I wish I could have found this earlier, doh. |
| 20:10 | <JonathanNeal> | But it was good of me to test too. |
| 20:14 | <SamB> | "I will *make* it legal" and all that, you know? |
| 20:56 | <TabAtkins> | SamB: Yeah, that's possible. I'd probably like to hold off on that until we have customizable combinators, and then just let people do that. |
| 20:56 | <TabAtkins> | a /_target/ * { box-shadow: 2px 2px black; } |
| 20:56 | <TabAtkins> | Where /_target/ is defined in JS. |
| 20:57 | <SamB> | hmm. |
| 20:57 | <SamB> | that would certainly be a *useful* thing to be able to do |
| 20:57 | <SamB> | though, working out a decent API to do this efficiently will be fun I bet |
| 20:58 | <SamB> | in particular, so that the CSS engine doesn't have to be calling the JS constantly |
| 21:00 | <TabAtkins> | Yeah, that's the rub. |
| 21:00 | <TabAtkins> | But I'm pretty sure it's solveable. |
| 21:00 | <Hixie> | // ftw |
| 21:01 | <Hixie> | though i didn't think of throwing JS into there |
| 21:01 | <Hixie> | that's an interesting idea |
| 21:01 | <TabAtkins> | /foo/ is the new "named combinator" syntax, which means it's now possible to plug JS in reasonably. |
| 21:02 | <SamB> | presumably, you'd want the JS code to be able to provide information about what it uses to do the computation |
| 21:02 | <TabAtkins> | The CSSWG accepted us adding in @selector-alias; to Selectors, and to pursue JS-driven version as well. |
| 21:02 | <TabAtkins> | That's a customizable pseudo-class, not combinator, but the problem space is basically identical. |
| 21:02 | <SamB> | so that it'd only need to do it again when some of that information changed |
| 21:02 | <TabAtkins> | Yup, there are various things we can do. |
| 21:03 | <TabAtkins> | Probably it'll come down to some fairly simple ways to say "don't call me unless [something] changes", where [something] is expressible as a Selector or some way to filter Object.observe records. |
| 21:12 | <SamB> | TabAtkins: so, like, a:hover[href] or so for this purpose? |
| 21:12 | <TabAtkins> | Yeah, something like that. |
| 21:13 | <SamB> | hmm, though of course if something mutated the href attribute that'd also change things ... |
| 21:15 | <SamB> | or messed around with id attributes such that the target element was a different element ... |
| 21:49 | <BS-Harou> | Can you think of any way to add feed from Opera 15+ extension to "Opera Mail" ? |
| 22:13 | <SimonSapin> | When does "&" actually need to be escaped in attribute values? (Context: http://www.databasesandlife.com/multilinelabelwithclickablelinks/ ) |
| 22:16 | <TabAtkins> | Whenever it looks like an escape. |
| 22:17 | <Hixie> | SimonSapin: for your sanity, escape it always |
| 22:17 | <TabAtkins> | So yes, always. |
| 22:17 | <Hixie> | SimonSapin: but yeah, in practice there's some complicated cases where it's ok not to escape it |
| 22:17 | <TabAtkins> | <div title="front¢er"> == oops |
| 22:17 | <Hixie> | e.g. href="?a=b&c=d" should be safe and won't trigger an error iirc |
| 22:17 | <Hixie> | but you're better off just always escaping it |
| 22:17 | <SimonSapin> | ok, thanks |
| 22:17 | <TabAtkins> | <div title="front¢er"> |
| 22:18 | <TabAtkins> | Hixie: That's safe? |
| 22:19 | <Hixie> | ¢er isn't safe, no, that one would be a conformance error |
| 22:19 | <TabAtkins> | Oh, it sounds like you said <a href="?front=foo¢er=bar"> would be safe. |
| 22:19 | <TabAtkins> | Were those a/b/c/d not metavars? |
| 22:20 | <Hixie> | i meant what i wrote literally, right |
| 22:20 | <TabAtkins> | Ah, kk. Gotta indicate that, it definitely looks like metavars. ^_^ |
| 22:20 | <Hixie> | that's why it's just simpler to always escape |
| 22:20 | <TabAtkins> | Yeah. |
| 22:20 | <Hixie> | i say what i mean and i mean what i say :-) |
| 22:21 | <TabAtkins> | The rule "it's safe to escape when you're not including anything that needs escaping". |
| 22:21 | <TabAtkins> | s/safe to escape/safe to not escape/ |
| 22:21 | <Hixie> | more or less |
| 22:22 | <Hixie> | the problem is that the definition of "anything that needs escaping" requires knowing a list of several thousand words :-) |
| 22:22 | <TabAtkins> | Yup yup, which is why it's a terrible rule. |
| 22:24 | <Hixie> | the reason for the rule is to avoid spurious error messages on pages that are fine in practice |
| 22:24 | <Hixie> | (it still catches cases that are likely broken because you aren't allowed to omit the semicolon in attributes) |
.png){kind=link}
