| 00:21 | <TabAtkins> | zewt: Just use xanthir.com/password for everything, remembering passwords is for chumps, as is counting on a password db that can be corrupted/doesn't follow you around. |
| 00:21 | <TabAtkins> | (You can save that to a local file, or put it up on your own server; it's an independent html file with no dependencies.) |
| 00:23 | <TabAtkins> | gsnedders: For no reason I can ascertain, one particular call of parseHTML (to read an <a><img></a>) is taking up 2.3% of Bikeshed's processing time. |
| 00:24 | <TabAtkins> | gsnedders: It's neither the first nor the last call to parseHTML; it's comfortably in the middle. |
| 00:24 | <TabAtkins> | I just can't figure it out. |
| 00:34 | <gsnedders> | TabAtkins: parseHTML? |
| 00:34 | <TabAtkins> | gsnedders: Sorry, it's an html5lib.parse() call. |
| 00:35 | <TabAtkins> | When I need to parse some independent html, I parse it as an independent document, then extract the contents of <body> in the returned document. |
| 00:36 | TabAtkins | is considering switching all of the include files to construct HTML elements directly, rather than being HTML text, but it's still confusing for now. |
| 00:37 | <TabAtkins> | Is there some sort of garbage collection that happens at some point? |
| 00:37 | <gsnedders> | nah |
| 00:37 | <TabAtkins> | It's possible I'm hitting that in Python itself, I suppose. |
| 00:38 | <TabAtkins> | It's just weird. I'm profiling Bikeshed, and I'm consistently getting 2-3% of processing time happening in addLogo(), which does nothing more than retrieve the logo.include file, parse it as HTML, then insert it. |
| 00:38 | <gsnedders> | what happens if you just profile that? |
| 00:38 | <TabAtkins> | I'm not doing line profiling yet. |
| 00:38 | <gsnedders> | where's the time spent within it? |
| 00:38 | <TabAtkins> | But I guess I can, hmm. |
| 00:39 | <gsnedders> | nah, I was more meaning just that callstack |
| 00:39 | <gsnedders> | so I guess call addLogo through cProfile.callctx or something |
| 00:39 | <TabAtkins> | Oh, the only part of the callstack taking up enough time to be worth showing in the profile is the call to parseHTML(). |
| 00:39 | <gsnedders> | Even just in that call to addLogo? |
| 00:39 | <TabAtkins> | Yes. |
| 00:40 | <gsnedders> | and it's internal to that function? |
| 00:40 | <TabAtkins> | That's literally the only thing showing up in the profile - the other calls are too small, are get pruned from the diagram. |
| 00:40 | <gsnedders> | could try explicitly calling gc.collect before, I guess |
| 00:41 | <TabAtkins> | As far as cProfile can tell, yeah, internal to that function. |
| 00:44 | <gsnedders> | could try asking in #python, idk any reason why html5lib would randomly take loads of time |
| 00:45 | <TabAtkins> | I'll try to dig down further when I'm done dealing with profiles here; as much as possible I'm trying to avoid calling html5lib anyway. |
| 00:45 | <TabAtkins> | It's a major part of Bikeshed's runtime. |
| 00:46 | <gsnedders> | well yes, parsers tend not to be quick in interpreted languages |
| 00:46 | <gsnedders> | too many function calls |
| 00:46 | <gsnedders> | and typically lots of copying of strings |
| 00:49 | <TabAtkins> | Right. I just didn't realize it would end up accounting for ~10% of my runtime. |
| 00:49 | <TabAtkins> | And by switching to direct lxml element creation, most of these functions would fall off of the profile entirely. |
| 00:50 | <TabAtkins> | Profiling has been very surprising and useful! |
| 00:50 | <gsnedders> | anolis was dominated by the cost of iterating of the tree, IIRC |
| 00:51 | <zewt> | tab: i'm sure there are lots of solutions--it just feels like something I shouldn't need an extra tool beyond the browser for by now |
| 00:53 | <TabAtkins> | gsnedders: Iterating only takes a small fraction of Bikeshed's time, interestingly - uses of the selector function are at about 6% of time. |
| 00:55 | <gsnedders> | TabAtkins: just getting all definitions and checking all possible refs took a large amount of time in anolis |
| 00:55 | <gsnedders> | TabAtkins: just because you have to go through so many elements |
| 00:55 | <TabAtkins> | That's interesting. I wonder why it took so long? |
| 00:57 | <TabAtkins> | I spend about 10% of my time initializing the reference manager, but actually looking up refs is only about 1.5% |
| 00:57 | <TabAtkins> | And that's using the full cross-ref database, across all the specs that Shepherd parses. |
| 00:59 | <gsnedders> | https://bitbucket.org/ms2ger/anolis/src/061718e61cb5bc196d1e805b1b5de720132380be/anolislib/processes/xref.py?at=default makes it look pretty complex deciding whether an element is suitable |
| 01:01 | <TabAtkins> | Interesting. |
| 01:12 | <TabAtkins> | gsnedders: Hmm, profiling another spec doesn't show the same pattern. Must just be an expensive GC that consistently triggers right there on that spec. |
| 01:15 | <gsnedders> | (Sorry, I'm a bit distracted by pointlessly reloading referendum results.) |
| 01:16 | <TabAtkins> | Haha, no problem. |
| 01:16 | <TabAtkins> | Only two counts in so far, dont' worry. |
| 01:16 | <TabAtkins> | Though fuck Orkney, for serious. |
| 01:17 | <TabAtkins> | It'll still be a few hours before results show up in several of the councils, anyway. |
| 01:17 | <gsnedders> | yeah, indeed |
| 01:17 | <gsnedders> | esp. for here |
| 01:17 | <gsnedders> | The islands will likely mostly vote no, none of that is surprising |
| 01:17 | <TabAtkins> | Yeah. |
| 01:18 | <TabAtkins> | They hardly give a shit about Westminster anyway. |
| 01:18 | <gsnedders> | The Islands feel abandoned by Westminster, mostly. But feel like they'd be more abandoned by a smaller country. |
| 01:18 | <gsnedders> | Instead of London being the centre of attenion, it'd be the Central Belt. |
| 01:19 | <TabAtkins> | Well, I'll get to see the results after dinner. |
| 01:19 | <TabAtkins> | Most councils will be reported in ~2 hours. |
| 01:20 | <gsnedders> | I should probably just sleep |
| 01:21 | <TabAtkins> | Yeah, wake up to a brand new day, or else Westminster madly backpedalling from their DevoMax promises. |
| 01:21 | <gsnedders> | I dunno, it's gonna be clear that a lot of Scotland is /really/ unhappy. I'm not sure doing nothing is politically tenable anymore. |
| 01:22 | <TabAtkins> | I honestly can't predict what the UK government is doing anymore. Y'all are schizophrenic right now with all the UKIP/Tory/etc business. |
| 01:24 | <gsnedders> | The problem is it's all FPTP and there's been no major change to let a third party get anywhere (see the rise of Labour and the decline of the Liberals), so once people get pissed off at the current government after n years here come the others |
| 01:26 | <TabAtkins> | Yeah, fptp is the worst voting system. It's not even "worst, except for all the others". It's just the worst. There's no excuse for it. |
| 01:33 | <gsnedders> | unfortunately it takes someone who cares more about constitutional reform than getting a majority in the house to actually change it |
| 01:36 | <gsnedders> | and also means getting the electorate over coalition governments |
| 01:38 | <TabAtkins> | Yeah, change is hard. :/ |
| 01:43 | roc | likes the NZ system, though we'll see how it works on Saturday |
| 01:43 | <roc> | TabAtkins: I wouldn't bother with Dr Olaf Hoffman, seriously |
| 01:44 | <gsnedders> | Like it's been very clear that people *do not like* coalition governments here. "But they're supporting polices that contradict their manifesto for governance!" |
| 01:46 | <roc> | I hope Scotland stays with the UK, because if they don't, we'll have to waste a lot of time and energy changing our flag |
| 01:47 | <gsnedders> | hah |
| 03:10 | <TabAtkins> | roc: Oh believe me, I know. Hoffman in an L2L thread, no less. But it still makes me angry to leave that sort of bullshit unchallenged. |
| 05:23 | <zcorpan> | Hixie: alt fix looks good, thanks |
| 07:44 | <MikeSmith> | annevk: in the list of params for the basic URL parser, as far the "optionally with an URL `url` and a state override `state override`" part, the optional `url` param there is a kind of spec-internal-only param that's just there as a hook for other parts of the algorithm, right? |
| 07:45 | <MikeSmith> | annevk: similar to the purpose of the `state override` param |
| 07:47 | <annevk> | MikeSmith: yeah isn't that mention in a note these days? |
| 07:47 | <MikeSmith> | annevk: dunno |
| 07:47 | MikeSmith | looks |
| 07:48 | <MikeSmith> | annevk: there's a note there that mentions other stuff but not this |
| 07:48 | <MikeSmith> | https://url.spec.whatwg.org/#concept-basic-url-parser |
| 07:49 | <MikeSmith> | oh wait |
| 07:49 | <annevk> | "The url and state override arguments are only for use by methods of objects implementing the URLUtils interface." |
| 07:49 | <MikeSmith> | "The url and state override arguments are only for use by methods of objects implementing the URLUtils interface. " |
| 07:49 | <MikeSmith> | yeah sorryt I missed that somehow |
| 07:58 | <MikeSmith> | annevk: there's a mozilla bug open for gecko not handling the host part of file: URLs per spec? |
| 07:59 | <MikeSmith> | I mean for a case like file://localhost/test it returns just "/test" |
| 08:12 | <annevk> | MikeSmith: I'm not sure |
| 08:12 | <annevk> | file URLs are generally problematic |
| 09:10 | <karlcow> | MikeSmith: do you mean file://localhost/test or file:///localhost/test |
| 09:11 | <karlcow> | aka scheme "file://" + full path "/localhost/test"; |
| 09:19 | <MikeSmith> | karlcow: two slashes only |
| 09:20 | <karlcow> | How would it work? if you mean relative link. |
| 09:20 | <MikeSmith> | karlcow: btw what's a good ramen place in Shimokita? |
| 09:21 | <karlcow> | houla… |
| 09:21 | <MikeSmith> | We're going to be there tonight |
| 09:21 | <karlcow> | I would rather go to Tobu Sakana (fish izakaya) than ramen but… hmmm ramen… |
| 09:22 | <karlcow> | there is one I like which has fish stew |
| 09:27 | <karlcow> | MikeSmith: |
| 09:27 | <karlcow> | 公式情報あり |
| 09:27 | <karlcow> | Mikesmith: 頭 ラーメン http://tabelog.com/tokyo/A1318/A131802/13045855/ |
| 09:29 | <karlcow> | but I prefer とぶさかな http://tabelog.com/tokyo/A1318/A131802/13001408/ |
| 11:25 | <annevk> | hsivonen: how much EV certificates is Mozilla paying for? Even mail.mozilla.org has one... |
| 11:26 | <annevk> | https://mail.mozilla.org/listinfo/ vs https://lists.mozilla.org/listinfo/ is so confusing |
| 12:10 | <mounir> | annevk: any comment on https://github.com/w3c/screen-orientation/pull/78 ? |
| 12:18 | <annevk> | mounir: it seems you'd want to change state in the same task, no? |
| 12:19 | <annevk> | mounir: but also, see comments from yesterday on hooks and animation frame task, for which I believe you already raised an issue with the web perf guys |
| 12:20 | <mounir> | annevk: the parts I'm changing are running in whatever task the page-visibility algorithm is running in |
| 12:20 | <annevk> | mounir: so should they be asynchronous? |
| 12:20 | <mounir> | annevk: there is no need to have this synchronous so async sounds good to me |
| 12:21 | <annevk> | mounir: aren't you dispatching events? |
| 12:21 | <mounir> | yes |
| 12:21 | <mounir> | in the screen orientation task though |
| 12:21 | <mounir> | that part is specified |
| 12:21 | <annevk> | mounir: so don't you want those to happen in the same task as the visibility events? |
| 12:22 | <mounir> | annevk: probably better to have all the screen orientation events fired in the screen orientation task? |
| 12:22 | <mounir> | annevk: maybe I could use the animation frame task when it's ready |
| 12:22 | <mounir> | but for the moment, using the same task for all those event firing seems the most sensible solution to me |
| 12:30 | <mounir> | annevk: ^ ? |
| 12:38 | <annevk> | hmm I guess |
| 12:38 | <annevk> | at this point I wonder how it's implemented |
| 12:38 | <annevk> | can't be this weird |
| 12:39 | <mounir> | annevk: in Chrome, the closest thing would be animation frame task |
| 13:03 | <hsivonen> | annevk: I have no idea of that cost info. |
| 13:03 | <hsivonen> | jgraham: sorry, I haven't had a chance to update the PR yet |
| 13:03 | <annevk> | hsivonen: what do you think about Safari only showing the EV name? |
| 13:04 | <annevk> | hsivonen: it's rather elegant on small screens |
| 13:05 | <hsivonen> | annevk: I don't have an iDevice, so I don't have enough data to comment |
| 13:13 | <jgraham> | hsivonen: np, you're just showing up as an overdue request in my bugzilla queue |
| 15:45 | <ericandrewlewis> | you know how when you're in an element with scroll bars, and when you get to the end of the scroll, the next scrollable element up the DOM chain immediately starts scrolling? |
| 15:45 | <ericandrewlewis> | is there any active discussion about how that UI should work? would that be at the HTML spec level, or is that an implementation detail for browsers to decide? |
| 15:48 | <TabAtkins> | ericandrewlewis: It's at the CSS spec level, or maybe DOM Events. |
| 15:48 | <TabAtkins> | But I dont' think scroll chaining has been properly defined anywher yet. |
| 15:49 | <ericandrewlewis> | cool, thanks TabAtkins. |
| 15:50 | <TabAtkins> | (I'd be interested in helping define that, for what it's worth.) |
| 15:53 | <ericandrewlewis> | TabAtkins: would a good first step be joining the mailing list, check for previous discussion, and starting conversation there? |
| 15:53 | <TabAtkins> | Yup! |
| 15:56 | <ericandrewlewis> | rad :) |
| 16:05 | <MikeSmith> | annevk: fwiw I think the org-name-only display that Safari does for sites with EV certs is great |
| 16:05 | <MikeSmith> | annevk: it seems fairly intuitive that if you want to see the URL you can just touch on the org name |
| 16:20 | <MikeSmith> | annevk: what Firefox Nightly does for http://people.w3.org/mike/phish/ is not so great through |
| 16:21 | <annevk> | MikeSmith: what does it do? |
| 16:21 | <MikeSmith> | hang on I'll make a screenshot |
| 16:26 | <MikeSmith> | annevk: https://gist.github.com/sideshowbarker/8284404#file-phish-png |
| 16:27 | <annevk> | wow |
| 16:27 | <annevk> | is that Android? |
| 16:27 | <annevk> | please file a bug |
| 16:27 | <MikeSmith> | yeah |
| 16:27 | <MikeSmith> | hai |
| 16:27 | <annevk> | ta |
| 16:53 | <doicia01_> | hi MikeSmith |
| 16:53 | <doicia01_> | are you in fact Michael[TM]Smith ? |
| 16:54 | <doicia01_> | test |
| 16:54 | <Andrei1000> | I was wondering about meta tags |
| 16:57 | <Andrei1000> | or meta-names rather |
| 16:57 | <Andrei1000> | https://help.whatwg.org/extensions/meta-name/ |
| 16:57 | <Andrei1000> | is an empty list! |
| 16:57 | <Andrei1000> | there's nothing there |
| 17:02 | <Andrei1000> | can anyone tell me why we're getting output such as: "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":698:43: error: Bad value �TYPEMETAL.FORMATPREFS�name�meta�typemetal.formatprefs� for all the meta-names listed in our checker/tests/html/elements/meta/names-registered-isvalid.html |
| 17:02 | <Andrei1000> | ? |
| 17:03 | <TabAtkins> | MikeSmith: ^^^ That's you, right? |
| 17:10 | <annevk> | Andrei1000: they're maintained on https://wiki.whatwg.org/wiki/MetaExtensions |
| 17:11 | <Andrei1000> | hi annevk ... it looks like the build.py |
| 17:11 | <MikeSmith> | TabAtkins: yea me |
| 17:11 | <Andrei1000> | doesn;'t like our checkertests/html/elements/meta/names-registered-isvalid.html file |
| 17:11 | <MikeSmith> | hmm yeah I need to fix that script I guess |
| 17:11 | <annevk> | Andrei1000: might be that things are failing due to recent move to using TLS |
| 17:11 | <MikeSmith> | seems like the move to TLS broke it |
| 17:12 | <MikeSmith> | what annevk said :) |
| 17:12 | <Andrei1000> | the public / w3 validator instance seems to be ok |
| 17:13 | <MikeSmith> | Andrei1000: yeah but it will break too next time I restart it |
| 17:13 | <Andrei1000> | that's exactly what happened to some of ours that were working ! :D |
| 17:13 | <Andrei1000> | but that explains a lot |
| 17:13 | <Andrei1000> | any idea when it will be fixed? |
| 17:14 | <annevk> | If it's written in Java that might be problematic, since hsivonen mentioned the other day that the Java module he uses doesn't support TLS SNI |
| 17:14 | <annevk> | If the script is not written in Java, it might be less bad |
| 17:14 | <MikeSmith> | annevk: it's a python script |
| 17:14 | <Andrei1000> | so you've now moved https://help.whatwg.org/extensions/meta-name/ from SSL to TLS? |
| 17:14 | <annevk> | MikeSmith: if it's Python 2.x, that again is likely problematic |
| 17:15 | <MikeSmith> | oh |
| 17:15 | <annevk> | Andrei1000: no, from not having encryption, to having encryption |
| 17:15 | <Andrei1000> | please restart the public/w3 validator instance, so people stop hassling us ;) |
| 17:15 | <MikeSmith> | it is Python 2 and running under Python 2.7 or whatever is installed by dreamhost |
| 17:16 | <MikeSmith> | Andrei1000: ? who's "us"? |
| 17:16 | <MikeSmith> | oh, and I missed the smile when I first read that |
| 17:16 | <Andrei1000> | we do QA stuff for BBC |
| 17:16 | <Andrei1000> | for their web portal |
| 17:16 | <MikeSmith> | ah ok |
| 17:17 | <Andrei1000> | they're saying it's our fault i.e. that our instances are broken, while the public instance is not |
| 17:17 | <Hixie_> | SSL to TLS? |
| 17:18 | <MikeSmith> | Andrei1000: tell them it's my fault :) and tell them to feel free to e-mail me at mike⊙wo if they want confirmation |
| 17:18 | <Andrei1000> | appreciated |
| 17:18 | <annevk> | MikeSmith: yeah only very latest Python 2.7 has a fix |
| 17:19 | <MikeSmith> | annevk: and dreamhost isn't running that? |
| 17:19 | <Andrei1000> | I'm going to hand you over to my collleague ... |
| 17:19 | <MikeSmith> | Andrei1000: OK, please do -- and I can send you a static file if you want |
| 17:19 | <jgraham> | You might be better off using requests for all your python http needs |
| 17:20 | <Andrei1000> | a static names-registered-isvalid.html you mean? |
| 17:21 | <Andrei1000> | MikeSmith: I don't get why the problem is the TLS on the remote site, when the build seems to be complaining of the local names-registered-isvalid.html .. |
| 17:21 | <annevk> | MikeSmith: DreamHost running the latest software? Hah |
| 17:21 | <annevk> | MikeSmith: "Python 2.6.6" is what DreamHost reports for me btw |
| 17:23 | <MikeSmith> | Andrei1000: the local validator build fetches https://help.whatwg.org/extensions/meta-name/ at build time and caches it locally |
| 17:24 | <MikeSmith> | every time your rebuild/restart it; e.g., if you do "python build/build.py all" |
| 17:25 | <Andrei1000> | MikeSmith: the one we're seeing locally looks perfectly readable ... but no the build doesn't like it |
| 17:25 | <Andrei1000> | still the build completes |
| 17:25 | <MikeSmith> | Andrei1000: hang on a minute there's to place you need to check |
| 17:26 | <Andrei1000> | I'll show you a sample of the o/p ... massive |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":622:34: error: Bad value �totalresults�name�meta�totalresults� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":623:34: error: Bad value �TOTALRESULTS�name�meta�totalresults� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":624:34: error: Bad value �ToTaLrEsUlTs�name�meta�totalresults� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":625:41: error: Bad value �twitter:app:country�name�meta�twitter:app:country� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":626:41: error: Bad value �TWITTER:APP:COUNTRY�name�meta�twitter:app:country� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":627:41: error: Bad value �TwItTeR:ApP:CoUnTrY�name�meta�twitter:app:country� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":628:47: error: Bad value �twitter:app:id:googleplay�name�meta�twitter:app:id:googleplay� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":629:47: error: Bad value �TWITTER:APP:ID:GOOGLEPLAY�name�meta�twitter:app:id:googleplay� |
| 17:26 | <Andrei1000> | "file:/opt/checker/tests/html/elements/meta/names-registered-isvalid.html":630:47: error: Bad value �TwItTeR:ApP:Id:gOoGlEpLaY�name�meta�twitter:app:id:googleplay� |
| 17:26 | <Andrei1000> | |
| 17:27 | <Andrei1000> | I think that's for all the entries in the file |
| 17:27 | <MikeSmith> | yeah that's expected if the https://help.whatwg.org/extensions/meta-name/ contents are empty :) |
| 17:27 | <MikeSmith> | and it's expected that the buildl will complete anyway if it's empty |
| 17:28 | <Andrei1000> | ok, good to know we're talking about the same thing :D |
| 17:30 | <NickHolmes> | MikeSmith, Hi. its me that's getting stress from our client. Do you know how we can resolve this issue by doing something to our instance of the validator to make this issue go away |
| 17:30 | <NickHolmes> | ? |
| 17:30 | <MikeSmith> | NickHolmes: yeah, first check the checker/validator/src/nu/validator/localentities/files/meta-name-extensions file in your build directory |
| 17:31 | <MikeSmith> | and don't run "build.py all" any more for a while |
| 17:32 | <NickHolmes> | MikeSmith: that file is empty |
| 17:32 | <NickHolmes> | but it it present |
| 17:32 | <NickHolmes> | it is |
| 17:32 | <TabAtkins> | It's the emptiness that's the problem. |
| 17:32 | <MikeSmith> | right |
| 17:33 | <NickHolmes> | so do we fill it with the contents of /checker/tests/html/elements/meta/names-registered-isvalid.html or something |
| 17:33 | <MikeSmith> | no. I can send you a non-empty up-to-date version of that file |
| 17:33 | <NickHolmes> | cool |
| 17:33 | <NickHolmes> | can you send it to andrei.doicin⊙rc please |
| 17:33 | <MikeSmith> | and you can drop it into that path and you should be ok |
| 17:33 | <MikeSmith> | ok |
| 17:33 | <annevk> | so I guess this is technically my fault, and I could blame it on Mike West for encouraging us to deploy TLS |
| 17:34 | <MikeSmith> | hah |
| 17:34 | <MikeSmith> | let's all blame Mike West |
| 17:34 | <MikeSmith> | blame chain |
| 17:34 | <annevk> | but this also seems good in a way, since getting TLS to work better is important |
| 17:34 | <NickHolmes> | MikeSmith: if we retstart, do we not overwrite it with emptiness? |
| 17:35 | <MikeSmith> | NickHolmes: it won't if you don't run "build.py all" but instead just do "build.py run" |
| 17:35 | <NickHolmes> | we will let you blame them, as long as its not our fault we dont care |
| 17:36 | <NickHolmes> | ok, can you send and we will try ... |
| 17:36 | <MikeSmith> | NickHolmes: you could avoid the need to do any building at all by running a release from https://github.com/validator/validator.github.io/releases/latest instead |
| 17:36 | <MikeSmith> | NickHolmes: yeah will send now |
| 17:36 | <NickHolmes> | cheers .... phew... I'm so looking forward to this being over... its taken 2 days to get to you, thinking it was something we did |
| 17:37 | <TabAtkins> | My Quest strikes again! |
| 17:38 | <MikeSmith> | NickHolmes: sorry this is all too arcane and not documented. Going forward I recommend running releases from https://github.com/validator/validator.github.io/releases/latest instead. |
| 17:41 | <MikeSmith> | NickHolmes: sent |
| 17:42 | <NickHolmes> | excellent - thanks |
| 17:42 | <TabAtkins> | MikeSmith: I recommend not writing anything that anyone else would ever want to use. That's the only way to prevent this from happening again. |
| 17:43 | <MikeSmith> | heh |
| 17:43 | <MikeSmith> | that's one way to do it |
| 17:43 | <MikeSmith> | NickHolmes: also posted to https://gist.githubusercontent.com/sideshowbarker/8284404/raw/5ddff4d08b7818c30645bf116c83aa43de2c9666/meta-name-extensions |
| 17:46 | <AndreiD> | thanks MikeSmith ... trying it shortly ... just to make sure again, we start the build.py like so: build.py run ... but what about that preset file? we normally run like so: |
| 17:46 | <AndreiD> | python ./build/build.py --follow-w3c-spec --presets-file="nu-validator-site/presets.txt" all |
| 17:47 | <AndreiD> | should we just change the "all" to "run" at the end? |
| 17:47 | <MikeSmith> | AndreiD: yes |
| 17:47 | <MikeSmith> | you don't need to change anything else |
| 17:49 | <MikeSmith> | AndreiD: ah wait sorry |
| 17:49 | <MikeSmith> | you need "build" too |
| 17:49 | <MikeSmith> | so: python ./build/build.py --follow-w3c-spec --presets-file="nu-validator-site/presets.txt" build run |
| 17:49 | <AndreiD> | gotcha |
| 17:57 | <SimonSapin> | zcorpan: Is the order of declarations in CSSStyleDeclaration defined? (For the purpose of .item(i)) |
| 17:59 | <TabAtkins> | SimonSapin: I think it's alphabetical? |
| 18:00 | <AndreiD> | MikeSmith: we tried it with build, and it blanks the meta-name-extensions file |
| 18:01 | <AndreiD> | we tried it without build i.e. just with run, and it gave us the same meta-name errors on the report |
| 18:01 | <MikeSmith> | AndreiD: gimme a minute |
| 18:03 | <MikeSmith> | AndreiD: try copying it to ./local-entities/meta-name-extensions as well |
| 18:05 | <annevk> | SimonSapin: it's IDL order |
| 18:05 | <AndreiD> | MikeSmith: should we still use build? |
| 18:05 | <MikeSmith> | AndreiD: yeah |
| 18:05 | <SimonSapin> | annevk: is that specified? |
| 18:10 | <MikeSmith> | annevk: btw about TLS browser UI/chrome I like that Google Chrome shows the URL with the "https" part in green, and like even better that desktop Safari shows the https only in a separate indicator (not as a part of the URL) with a gray background (instead of white) together with the lock icon |
| 18:12 | <annevk> | SimonSapin: I think it was at one point |
| 18:12 | <annevk> | MikeSmith: I would like not showing "https" at all since it's a confusing string |
| 18:13 | <MikeSmith> | annevk: long term I guess that would be better |
| 18:13 | <MikeSmith> | short term you need to deal with current user expectations I would think, and not disrupt them |
| 18:14 | <MikeSmith> | and with current third-party guidance to users about what to look for in order to make trust decisions |
| 18:16 | <MikeSmith> | annevk: I think you wrote that URLs for TLS should be shorter not longer, and that makes sense |
| 18:17 | <MikeSmith> | we really should flip it around some and show a warning indicator for non-TLS URLs instead |
| 18:17 | <annevk> | yeah, starting with not making the TLS UI worse would be good |
| 18:19 | <Hixie_> | i proposed that like 10 years ago :-) |
| 18:20 | <MikeSmith> | the Web always learns thing the hard way |
| 18:21 | <jgraham> | MikeSmith: I think Opera's approach here is pretty good |
| 18:21 | <MikeSmith> | btw as bad as the Web platform is, it at least doesn't have anything as fragile as BGP |
| 18:23 | <MikeSmith> | the http://www.renesys.com/2014/09/why-the-internet-broke-today/ thing |
| 18:23 | <annevk> | jgraham: yeah, as is mobile Safari |
| 18:23 | <MikeSmith> | "such is the nature of our trust-based Internet routing. Pretty much anyone can mess it up" |
| 18:24 | <jgraham> | annevk: I will verify that once they release a version for FirefoxOS :p |
| 18:27 | <MikeSmith> | jgraham: oh, Opera never shows "http" or "https" in the address bar at all? just the lock for TLS and lock + org name for EV? (and just the non-scheme part of the URL for non-TLS?) |
| 18:29 | <jgraham> | MikeSmith: It does if you focus the address bar (so that copy and paste works in a simple way), but otherwise a label like "Secure" or "Web" (maybe they changed the exact words now) slides over the scheme |
| 18:30 | <MikeSmith> | ah ok |
| 18:30 | <MikeSmith> | yeah, that's pretty nice |
| 18:32 | <Philip`> | MikeSmith: Fortunately BGP was designed so that at most 65536 different people had to be trusted not to mess it up (due to using 16-bit AS numbers) |
| 18:32 | <Hixie_> | yeah how's that worked out |
| 18:32 | <Hixie_> | it's not like india has ever blackholed the entire world's youtube traffic or anything like that |
| 18:33 | <Philip`> | MikeSmith: but unfortunately someone decided it was a good idea to extend it to 32-bit AS numbers (which will surely be big enough forever) |
| 18:33 | <Hixie_> | (or was it pakistan, i forget) |
| 18:33 | <Philip`> | MikeSmith: (Also unfortunately, 32-bit AS numbers actually broke the internet, like in http://www.renesys.com/2009/05/byte-me/ ) |
| 18:34 | <jgraham> | MikeSmith: https://www.youtube.com/watch?v=OKcVXbLaHN0 seems to be a rather dated and inaccurate video about it |
| 18:36 | <MikeSmith> | Philip`: hadn't read that one yet |
| 18:41 | <MikeSmith> | jgraham: so yeah I see that current opera seems to have moved beyond that a bit and doesn't show the scheme even when you focus address bar. You can't see it even when you select it -- only after you paste it |
| 18:41 | <MikeSmith> | as far as I can tell |
| 18:42 | <MikeSmith> | anyway it's a pretty clean design |
| 18:43 | <jgraham> | Oh, that doesn't really sound better :) |
| 18:43 | <MikeSmith> | jgraham: sounds more like what annevk wants at least |
| 18:44 | <MikeSmith> | Hixie_: yeah that was Pakistan it seems http://www.renesys.com/2008/02/pakistan-hijacks-youtube-1/ |
| 18:44 | <jgraham> | copy and paste adding random things you didn't actually select to the output seems pretty awful |
| 18:44 | <MikeSmith> | gotta love that Renesys blog |
| 18:45 | <MikeSmith> | jgraham: Chrome does it too I think |
| 18:45 | <MikeSmith> | or wait no Firefox does |
| 18:45 | <MikeSmith> | for http URLs |
| 18:46 | <MikeSmith> | it doesn't show the "http" even when you focus in the address bar |
| 18:46 | <MikeSmith> | and even when yuo select |
| 18:46 | <MikeSmith> | ah yeah and neither does Chrome -- they both only show it for https URLs |
| 18:47 | <jgraham> | That doesn't mean it's good |
| 18:47 | <MikeSmith> | I guess this is maybe part of what annevk was talking about in his blog post |
| 18:48 | <MikeSmith> | jgraham: I didn't say it meant it was necessarily good :) |
| 18:48 | <MikeSmith> | though I personally don't think it's awful |
| 18:50 | <MikeSmith> | AndreiD: I need to get some sleep. So I hope you got you local validator instance working OK with that file I sent. If not, I'll be around again in a few hours and will try to figure out how else to fixe it |
| 18:50 | <jgraham> | I don't know about platforms that are actually supported, but Opera 25/Linux seems to be considerably worse than 12 |
| 18:50 | <jgraham> | In this respect |
| 18:58 | <Philip`> | MikeSmith: By "love", do you mean "be terrified by"? |
| 19:09 | <zcorpan> | SimonSapin: yes |
| 19:09 | <Hixie_> | MikeSmith: sounds right |
| 19:09 | <SimonSapin> | zcorpan: where is it defined? |
| 19:11 | <zcorpan> | SimonSapin: look at the things that return cssstyledeclaration |
| 19:12 | <zcorpan> | canonical order or so i think for .style |
| 19:14 | <SimonSapin> | zcorpan: http://dev.w3.org/csswg/cssom/#the-elementcssinlinestyle-interface says specified (source) order |
| 19:15 | <Hixie_> | TabAtkins: you around? |
| 19:19 | <Hixie_> | TabAtkins: 'font' doesn't seem to have a matching <font> for me to parse as |
| 19:21 | <Hixie_> | TabAtkins: see https://www.w3.org/Bugs/Public/show_bug.cgi?id=26208 |
| 19:45 | <terinjokes> | paul_irish: officially issued a ticket |
| 19:45 | <terinjokes> | (y) |
| 19:53 | <TabAtkins> | Hixie_: Which part of <font>? |
| 19:54 | <Hixie_> | the 'font' shorthand property's grammar |
| 19:54 | <Hixie_> | see the bug |
| 19:54 | <TabAtkins> | Oh, I see. |
| 20:07 | <JakeA> | marcosc: how does one correctly pronounce your surname? |
| 20:07 | <TabAtkins> | Caceres? |
| 20:07 | <TabAtkins> | Kah-sair-ehz |
| 20:07 | <TabAtkins> | (I'm crap at spelling pronunciation, apparently.) |
| 20:08 | <TabAtkins> | That's assuming standard spanish pronunciation rules. |
| 20:08 | <JakeA> | I might just make it up |
| 20:08 | <TabAtkins> | Call me, I'll pronounce it for you. |
| 20:10 | <JakeA> | TabAtkins: Thanks! |
| 20:10 | <JakeA> | I may use that service again |
| 20:10 | <TabAtkins> | Hahahaha |
| 20:10 | <TabAtkins> | Why do you need it? |
| 20:10 | <TabAtkins> | s/do/did/ |
| 20:11 | <JakeA> | Moderating a panel at Edge tomorrow |
| 20:11 | <TabAtkins> | Oh, yeah, makes sense. |
| 20:12 | <annevk> | MikeSmith: I think if you focus the address bar it makes some sense to reveal the full URL |
| 20:13 | <annevk> | MikeSmith: but when you're viewing a site and glance at the address bar it should be immediately obvious a) where you are and b) whether it's secure (or the reverse, whether it's insecure) |
| 20:13 | <TabAtkins> | Agree - I like the trend of only showing the domain unless you're focusing it. Makes it harder to phish. |
| 20:14 | <TabAtkins> | evil.cc/paypal.com |
| 20:16 | <JakeA> | TabAtkins: yeah, but did you see the comments on my blog post about that? |
| 20:16 | <TabAtkins> | No, I didn't. Link? |
| 20:16 | <annevk> | "Blink made hz-gb-2312 as a replacement encoding." <<< hsivonen |
| 20:16 | <JakeA> | http://jakearchibald.com/2014/improving-the-url-bar/ |
| 20:16 | <annevk> | -- https://www.w3.org/Bugs/Public/show_bug.cgi?id=25339#c4 |
| 20:16 | <JakeA> | At one point wife-beating comes up |
| 20:17 | <annevk> | JakeA: poor JakeA, advocating TLS while not using it :p |
| 20:17 | <JakeA> | (I did get a bit agitated in the comments from all the conspiracy theory) |
| 20:17 | <JakeA> | hah, yeah |
| 20:17 | <JakeA> | I know I know |
| 20:17 | <terinjokes> | JakeA: :( come on, SSL is easy |
| 20:18 | <JakeA> | I hate you all |
| 20:18 | <JakeA> | It's on my TODO list |
| 20:19 | <JakeA> | Although it keeps getting bumped by things like "play computer games until vomit" |
| 20:20 | <Hixie_> | spec.whatwg.org should work now btw |
| 20:20 | <Hixie_> | (no thanks to dreamhost) |
| 20:20 | <terinjokes> | JakeA: :( |
| 20:21 | <annevk> | Hixie_: do we know what was up? |
| 20:21 | <Hixie_> | annevk: their panel is buggy |
| 20:22 | <Hixie_> | annevk: you have to first set up the http: site, then set up the https: site, then you can edit the http: site |
| 20:22 | <annevk> | Hixie_: it points to a Polish translation that is 410'd |
| 20:22 | <annevk> | Hixie_: o_O |
| 20:22 | <Hixie_> | if you set up the https: site without first setting up the http: site, it just does whatever the http: site would do |
| 20:22 | <Hixie_> | wait, what? |
| 20:22 | <Hixie_> | oh the link at the top |
| 20:23 | <Hixie_> | removd |
| 20:24 | <Domenic> | Ooh and wiki stylesheet is fixed too |
| 20:24 | <Domenic> | I think we may have reached the point of a successful transition |
| 20:24 | <Domenic> | Wait! Mixed content on https://developers.whatwg.org/ |
| 20:25 | <Domenic> | http://fonts.googleapis.com/css?family=Droid+Serif:regular,italic,bold |
| 20:25 | <Domenic> | Hmm and stylesheet broken on https://wiki.whatwg.org/wiki/What_you_can_do O_O |
| 20:41 | <Hixie_> | Domenic: developers.whatwg.org is going to be broken for some time still |
| 20:42 | <Hixie_> | since i need to write the code to regenerate it |
| 20:42 | <Domenic> | Hixie_: yeah, that is a sad |
| 20:42 | <Hixie_> | for hte wiki, it's a caching issue |
| 20:42 | <Hixie_> | logged-out users get cached pages |
| 20:42 | <Hixie_> | that were cached before hte transition |
| 20:43 | <Hixie_> | dbaron, TabAtkins: ping https://www.w3.org/Bugs/Public/show_bug.cgi?id=26264 |
| 20:46 | <dbaron> | ugh, line breaking rules |
| 20:46 | <TabAtkins> | dbaron: What are your thoughts on doing more of those things within the confines of 'display'? |
| 20:47 | <dbaron> | TabAtkins, I guess I probably prefer display to ::after + content, although I wonder how important wbr is... |
| 20:47 | <TabAtkins> | dbaron: I use it in Bikeshed. ^_^ |
| 20:48 | <TabAtkins> | (Insert it around the em-dashes it typography-fixes into the doc, since there aren't any spaces.) |
| 20:48 | <Hixie_> | dbaron: we ended up making <wbr> conforming due to the amount of usage it has |
| 20:48 | <Hixie_> | dbaron: so... |
| 20:49 | <dbaron> | TabAtkins, as a substitution for --- ? |
| 20:49 | <TabAtkins> | dbaron: If you end a line with --, yeah, it turns it into an em-dash. |
| 20:50 | <dbaron> | TabAtkins, hmmm, the TeX convention is -- for an en-dash, and --- for an em-dash |
| 20:50 | <TabAtkins> | I didn't give a lot of thought to it, but I bet I could find all the uses of -- in the CSSWG and just fix them. |
| 20:50 | <dbaron> | I guess I'm not as concerned about the performance of <wbr> as I am of <br> |
| 20:51 | <TabAtkins> | If we do <br> as display, though, we should do <wbr> too. |
| 20:54 | <TabAtkins> | Hixie_: Re: font, is it just the fact that the global keywords might be included in the <'font'> nonterminal? |
| 20:54 | <Hixie_> | TabAtkins: what is the <'font'> nonterminal? |
| 20:55 | <Hixie_> | i looked up 'font' in the CSS Fonts spec and it didn't have a single grammar token for me to reference |
| 20:55 | <TabAtkins> | The non-terminal that stands for "the value of the 'font' property" |
| 20:55 | <TabAtkins> | <'foo'> in general means that. |
| 20:55 | <Hixie_> | oh |
| 20:55 | <Hixie_> | where is that defined? |
| 20:55 | <Hixie_> | and does it include 'inherit' and so on? |
| 20:55 | <TabAtkins> | To your first question, http://dev.w3.org/csswg/css-values/#component-types |
| 20:56 | <TabAtkins> | To your second, no. |
| 20:56 | <TabAtkins> | (Also defined in that link.) |
| 20:56 | <Hixie_> | if it doesn't incldue the global keywords then it sounds perfect |
| 20:56 | <TabAtkins> | It is exactly what you need, yes. |
| 20:57 | <Hixie_> | coolio |
| 20:57 | <Hixie_> | thanks |
| 20:57 | <TabAtkins> | You're welcome. In return, you have to never say "coolio" again. |
| 21:01 | <Hixie_> | what's wrong with coolio |
| 21:05 | <TabAtkins> | Everything. |
| 21:36 | <annevk> | 'what's with these @w3c specs having generic, grandiose & non-descript titles? "Encoding", "URL". what's next? "Web"?' |
| 21:37 | <Hixie_> | i wish |
| 21:37 | <annevk> | can we uncache the wiki? |
| 21:37 | <annevk> | where's gphemsley? |
| 22:17 | <Domenic> | wow I did not know about <br clear> |
| 23:45 | <Hixie_> | terinjokes: https://www.w3.org/Bugs/Public/show_bug.cgi?id=26182 |
| 23:52 | <terinjokes> | Hixie_: i've been chating with brian |
| 23:52 | <terinjokes> | i don't have an account t comment on it |
| 23:52 | <Hixie_> | easy to get one :-) |
| 23:52 | <terinjokes> | oh, ok |
| 23:52 | <Hixie_> | anyway, i'm leaving this in your court now |
| 23:52 | <terinjokes> | if you insist :P |
| 23:52 | <Hixie_> | you don't have to |
| 23:52 | <Hixie_> | entirely up to you :-) |
| 23:55 | <terinjokes> | oh, i just made the connection between the commenter and you |
| 23:58 | <Domenic> | haha ^_^ |