| 00:15 | <Hixie_> | GPHemsley: yeah unfortunately STS requires that the top-level domain be a superdomain of other subdomains |
| 00:15 | <Hixie_> | GPHemsley: so www. had to go |
| 00:18 | <Hixie_> | GPHemsley: <code> should be orange if it's a xref |
| 00:49 | <tantek> | "mailing lists are the worst form of communication invented" - @slightlylate |
| 00:50 | <tantek> | good to see the message spreading |
| 00:50 | <tantek> | "mailing lists … recipe for failure" |
| 00:55 | <Hixie_> | yeah, worst except for all the others :-) |
| 00:57 | <tantek> | Hixie, except IRC :) |
| 00:57 | <Hixie_> | for technical discussion, IRC is almost as bad as twitter :-P |
| 00:57 | <tantek> | your usage is different - it's your inbox :) |
| 00:57 | <tantek> | Hixie - no way - the level of conversation in here is typically 10x better than Twitter # madeupstats |
| 01:12 | <tantek> | lol "for analytics" |
| 01:25 | <tantek> | go sicking! loving all this open questioning of the extensible web manifesto. |
| 01:30 | <tantek> | lol @slightlyoff "just need to put a web server on the thumb drive" |
| 01:31 | <Phae> | He's on top form today |
| 01:33 | <sicking> | tantek: i wouldn't say that i'm questioning the extensible web manifesto |
| 01:33 | <sicking> | tantek: just saying that it's not always possible to do. |
| 01:34 | <tantek> | sicking - I'd say your point about device APIs outright questioned the prioritization of the extensible web manifesto |
| 01:34 | <sicking> | tantek: in the context of security sensitive APIs I don't think it's always possible to do |
| 01:36 | <tantek> | more like nearly never possible to do. |
| 01:40 | <Hixie_> | tantek: dare i ask where you're seeing this conversation? |
| 01:43 | <Phae> | He'll be watching the edgeconf love feed |
| 01:43 | <Phae> | Live* even. |
| 01:43 | <Phae> | At Adobe right now |
| 01:45 | <tantek> | Hixie: edgeconf.com for most of today. just wrapped up. |
| 01:45 | <tantek> | Hixie: I just presumed #whatwg as the default backchannel for all things web platform ;) |
| 01:46 | <Hixie_> | oh this is a talk where alex and sicking are saying this stuff? |
| 01:46 | <tantek> | right, conference of panels, some folks on the panel, some in the audience |
| 01:46 | <tantek> | quite a solid gathering |
| 02:14 | <MikeSmith> | if mailing lists are the worst form of communication invented, then I guess we need to mint a new word stronger than "worst" to describe telephone conferences |
| 05:38 | <GPHemsley> | I think telegraph conversations are worser |
| 05:44 | <terinjokes> | any other ham radio guys around… |
| 12:20 | <annevk> | https://store.apple.com redirects to http:// ... |
| 13:33 | <MikeSmith> | AndreiD: fyi I fixed https://help.whatwg.org/extensions/meta-name/ |
| 13:34 | <MikeSmith> | AndreiD: so you guys can switch your validator build invocation back to "python ./build/build.py ... all" safely |
| 13:35 | <MikeSmith> | annevk: cool, just like the NSA does (https redirect) |
| 13:35 | <MikeSmith> | annevk: so the python issue is that libxml2 in python <2.7 doesn't handle https? |
| 14:14 | <GPHemsley> | Why is anyone still using Python <2.7? |
| 14:17 | <Ms2ger> | debian-stale? |
| 14:17 | Ms2ger | has no idea |
| 14:23 | <MikeSmith> | dreamhost |
| 14:23 | <MikeSmith> | debian stable has had 2.7 for a long time, I think |
| 14:25 | <MikeSmith> | so I guess the next question is why is anyone still using dreamhost |
| 14:26 | <MikeSmith> | dreamhost is the sourceforge of .. something |
| 14:27 | <MikeSmith> | somebody please give me a heads-up when tantek catches the TLS-only fever |
| 14:28 | <MikeSmith> | so I can switch my scraper for http://microformats.org/wiki/existing-rel-values |
| 14:53 | <annevk> | MikeSmith: I tried to find exact details the other day but couldn't |
| 14:53 | <annevk> | MikeSmith: but I think it's Python 2.7.9 or some such |
| 14:53 | <annevk> | MikeSmith: http://legacy.python.org/dev/peps/pep-0466/ is related |
| 14:53 | <annevk> | MikeSmith: I think it was planned for 2.7.7 but did not make it there |
| 14:55 | <annevk> | Seems 2.7.9 is not yet out |
| 14:58 | <annevk> | MikeSmith: and http://bugs.python.org/issue5639#msg204323 |
| 14:58 | <annevk> | MikeSmith: mnot commented there as well |
| 14:59 | <annevk> | MikeSmith: found https://docs.python.org/2/library/ssl.html#ssl.HAS_SNI |
| 14:59 | <annevk> | MikeSmith: suggests it's 2.7.9 which as far as I can tell is not even released yet? |
| 15:01 | <MikeSmith> | annevk: ah Ok it might not actually be the cause of problem I'm seeing, I dunno |
| 15:03 | <annevk> | MikeSmith: I think there's a way to use some other package to get SNI, but it has a ton of dependencies |
| 15:03 | <MikeSmith> | annevk: what I was using is lxml.html.parse and that apparently can't handle https URLs |
| 15:03 | <MikeSmith> | annevk: OK |
| 15:03 | <annevk> | MikeSmith: handle them at all? |
| 15:04 | <MikeSmith> | annevk: yeah based on glancing through some stuff on StackOverflow, I think not at all |
| 15:04 | <MikeSmith> | error logs don't provide much detail |
| 15:04 | <annevk> | MikeSmith: you could try perhaps with a resource that has a dedicated IP address |
| 15:04 | <annevk> | MikeSmith: maybe google.com? |
| 15:04 | <annevk> | (not sure if Google has restrictions in place, I'd guess they do) |
| 15:05 | <MikeSmith> | ah OK I guess I cuold try that |
| 15:06 | <MikeSmith> | annevk: anyway I worked around it for now by instead just using urllib2.urlopen to get a file object to read from, and then using lxml.html.document_fromstring to get a tree from that |
| 16:06 | <annevk> | https://twitter.com/janl/status/513714191665233920 |
| 16:06 | <annevk> | MikeSmith: I guess urllib2 might be the one with all the dependencies then if that just works? |
| 16:07 | <annevk> | MikeSmith: seems fine then, I also mostly use urllib2 iirc |
| 16:08 | <GPHemsley> | annevk: FYI, I've returned to active duty as the mimesniff editor, if you need to coordinate things. (It seems there have been developments in Fetch that could use corresponding support in mimesniff.) |
| 16:10 | <GPHemsley> | MikeSmith: And what is the github equivalent of Dreamhost? |
| 16:11 | <GPHemsley> | DreamHost : SourceForge :: ??? : Github |
| 16:16 | <Ms2ger> | GPHemsley, your next business ;) |
| 16:16 | <GPHemsley> | pfft |
| 16:16 | <GPHemsley> | no thanks |
| 16:22 | <MikeSmith> | I guess the github equivalent of Dreamhost is hosting on your own virtual machine that you can actually control all the software on |
| 16:24 | <annevk> | GPHemsley: great to hear |
| 16:24 | <annevk> | GPHemsley: it seems working on the nosniff stuff could help Gecko |
| 16:25 | <annevk> | GPHemsley: I think mimesniff is prolly mostly used after Fetch has retrieved a resource, though I could envision some convenience spec hooks that couple them |
| 16:36 | <GPHemsley> | annevk: It's not clear to me what the issues are with nosniff as I've documented it. |
| 16:36 | <GPHemsley> | (If any) |
| 16:41 | <GPHemsley> | annevk: Unless you're referring to expanding the applicability of the image-type sniffing |
| 16:41 | <GPHemsley> | to additional contexts |
| 21:58 | GPHemsley | glares at annevk for using relative dates on all his blog posts |
| 21:59 | <caitp> | relative times are nice to look at |
| 21:59 | <tantek> | GPHemsley you may find this interesting: http://aaronparecki.com/articles/2012/08/23/1/you-should-not-be-displaying-relative-dates |
| 22:00 | <caitp> | exact times are nice as alt text |
| 22:01 | <caitp> | "relative dates break screenshots in a universe where exif doesn't exist" |
| 22:01 | <tantek> | caitp - welcome to that universe |
| 22:01 | <caitp> | i'm not in that universe really |
| 22:02 | <tantek> | screenshots on iOS and MacOS are PNGs - without any EXIF |
| 22:02 | <GPHemsley> | So, annevk uses <data> but not @title |
| 22:02 | <caitp> | sure, but that's a fixable problem |
| 22:02 | <GPHemsley> | and by <data> I mean <time> |
| 22:02 | <tantek> | so, which universe/platform are you on which puts EXIF in screentshots? |
| 22:02 | <caitp> | browsers can do it, WMs can do it |
| 22:02 | GPHemsley | consistently forgets that the element is <time> and not <date> |
| 22:03 | <tantek> | caitp - did't ask who "can" - asked who *does* |
| 22:03 | <tantek> | feel free to demonstrate EXIF in a screenshot by posting a URL to an example. |
| 22:03 | <caitp> | who cares :p the point is all about the fact that metadata exists |
| 22:04 | <tantek> | have yet to see *any* platform/device actually put EXIF in screenshots |
| 22:04 | <caitp> | so resolving the absolute time from relative times in a screenshot is possible |
| 22:04 | <tantek> | nope, it doesn't exist - you're making it up ;) |
| 22:04 | <tantek> | i.e. citation needed |
| 22:04 | <caitp> | we know that image metadata does exist, it's used in basically every cellphone and digital camera ever |
| 22:04 | <caitp> | whether it's used by OSes is irrelevant |
| 22:04 | <caitp> | it could be used |
| 22:04 | <caitp> | if it's not used, it's arguably a bug |
| 22:05 | <caitp> | people deal with relative time better than exact dates |
| 22:05 | <caitp> | it's harder to remember what you were doing on 9/08/2014 than it is to remember what you were doing 2 weeks ago |
| 22:06 | <caitp> | i mean, I don't have data to support that claim, it's anecdotal from my own experience |
| 22:06 | <caitp> | but if the data were collected, you'd probably find that |
| 22:07 | <GPHemsley> | oh, great, and he uses "fortnight" |
| 22:16 | GPHemsley | looks at OpenSSL and wonders where Poerto Rico is |
| 22:17 | <GPHemsley> | "HTTP 405 - Method Not Allowed |
| 22:17 | <GPHemsley> | (If you are a human being, this should have never happened)" |
| 22:17 | <GPHemsley> | Hmm |
| 22:17 | GPHemsley | reevaluates himself |
| 22:19 | <GPHemsley> | oh, I guess I'm back to being a human now |
| 22:19 | <tantek> | caitp - yes it's (arguably) a bug - OS screenshots don't save EXIFs, which last time I checked, only apply to JPEGs, whereas screenshots use PNG typically. |
| 22:20 | <caitp> | or bmp if you're super lucky! |