2015-10-07 UTC
nitot and KartikPrabhu joined the channel
nitot joined the channel
# 01:20 GWG tantek: Does that explain the issue I was talking about?
# 01:21 tantek GWG, not quite - I'm hoping you can provide specific examples?
# 01:21 tantek the challenge here is that theoretically I can imagine how any page *could* have a reasonable h-feed
# 01:21 tantek e.g. a post permalink could contain a nested h-feed of the responses to it!
# 01:22 tantek nearly any page could have a nested h-feed of the revisions to the page!
# 01:22 GWG tantek: But this is a top-level h-feed
# 01:23 tantek the point is that simply the presence of an h-feed is not bad nor a problem per se
# 01:25 tantek The problem is that when you ask what is the problem, you need to be more specific.
# 01:26 tantek GWG, what specific problem are you encountering with a top level h-feed?
# 01:28 GWG tantek: My objection is that people are just putting it in without thought. It is a decision become copied over and over again.
# 01:29 tantek sounds like something we can perhaps suggest for indiewebify.me
# 01:29 GWG tantek: This is the same decision in WordPress that has hentry automatically added to everything. The implementation was wrong when they did it, and it has just propogated.
# 01:29 tantek that is - noticing that there is an extraneous h-feed, and reporting it as an error
# 01:30 tantek GWG, but that decision is not completely wrong either - h-entry is a very good default top level microformat for a page
# 01:31 GWG WordPress uses two class functions to automatically insert classes...body_class and post_class.
# 01:31 tantek GWG, I think the real problem here is that the semantic markup is being split across core and templates
# 01:32 GWG tantek: What's the solution for that?
# 01:32 tantek whereas it really should be all in one or the other
# 01:32 GWG Templating makes sense in some cases.
# 01:32 tantek because when it is split across core and templates, it's nearly impossible for EITHER side to fix it!
# 01:32 tantek GWG, nah, if you can't put all of the default markup in core, then it should all be put into templates
# 01:33 tantek this may be a WordPress architectural issue - the fact that core has a body_class and post_class at all is likely a mistake
# 01:33 tantek whereas the solution would be to empower a template to control the entire page markup
# 01:34 tantek and then different kinds of templates for different pages etc.
wolftune joined the channel
# 01:34 tantek one should be able to completely change the markup without touching core code
# 01:34 tantek so this is a bigger WordPress architectural problem than just "too many hfeeds or too many hentrys"
# 01:35 tantek happy to have snarfed, or any other WordPress users here review my reasoning / find any flaws.
# 01:35 GWG But every time I talk to the themers, they say it is a core issue.
# 01:35 GWG The core people say it is a theme issue
# 01:48 tantek You can point out that it's a core architectural issues, because core should not be interfering with markup - that's the theme's job and responsibility.
# 01:48 tantek The themers are right. When the core people say it is a theme issue - ask them, how do they propose a theme "undo" the markup that is hard coded in core?
# 01:56 tantek if we can't fix WordPress core, at least we can document the architectural error so that other implementations don't make the same mistake
# 01:57 tantek and maybe someday someone will be able to fix WordPress core, maybe they even have an outstanding issue on this!
# 02:02 GWG tantek: I've been working on other issues. I just closed a 9 year old issue for them about pingbacks.
# 02:02 tantek GWG, great - please add citations to the issues you file!
nitot joined the channel
wolftune and KevinMarks joined the channel
# 02:38 KevinMarks GWG I love that you closed a 9-year old bug that references a dead site
# 02:40 GWG KevinMarks: This is the WordPress Pingback code. It really needs a massive update.
# 02:44 GWG If only I was better at parsing. I always had issues with that.
shaunguice joined the channel
# 02:56 tantek massive FB invitations for tomorrow's HWC SF finally sent out
lukebrooker joined the channel
wolftune joined the channel
# 03:05 Loqi [mention] Kevin Smokler, Ben Moskowitz, George Kelly, Colleen Taylor, Atul Varma, Lawrence Lessig, Cariwyl Hebert, David Baron, Jay Zalowitz, Justin David Kruger, Stephanie Vacher, Stacey Rivet, Marie Williams, Jeff Hodsdon, Jeffrey Veen, Megs ORorke, Erin Stevenson O'Connor, Olya Lapina, Daniel Appelquist, Evan Prodromou, Chris Heuer, Stephen Wyatt Bush, James Craig, Tony Rai, Michael Owens, Joël Franusic, Kara Murphy, Loic Le Meur, Amy MacKinnon, Zibi Braniecki, Jeff Rider, Thomas Vander Wal, Kitt Hodsden, Jesse Vincent, C Alaric Moore, Stephanie Haupt Sullivan Rewis, Simon Law, Zoe Schiffer, Brian Behlendorf, Martin Atkins, Christine Herron, Paul Hammond, Beau Smith, Charles Hope, Jen Bradburn, Matthew Levine, Christopher Carfi, Mo Kudeki, Peter Hirshberg, Lizz Noonan, Nima Dilmaghani, Om Malik, Cari Levay, Laura Helen Winn, Pius Uzamere, Jeremy Anderson, Matt Schaar, Carla Borsoi, Joichi Ito, Jordan McArthur, Thor Muller, Brynn Evans, Yan XZ, Jessica Suttles, Ben Metcalfe, Adam Rifkin, Nate Koechley, Mar
nitot joined the channel
# 03:33 aaronpk good thing we archive all that content to our wiki regularly
benwerd joined the channel
# 03:36 aaronpk looks like they're making the plaintext content available at a new URL base
# 03:55 wagle ok, they explain it, but i'm reading fast
# 03:56 wagle i bring up microdata since one of my projects here is grokking it, where-ever it fits into the puzzle
# 03:58 wagle but you might be forgiven for assuming that since they bring it up in week 1 of the mooc
wolftune joined the channel
# 03:58 aaronpk yup that's kind of unfortunately the problem with that course
todrobbins joined the channel
# 04:05 wagle well, i'm pretty ignorant of the whole html/css/whatever-version-of-out-of-band-data-you-are-using, so I hope learning about microdata transfers to others
# 04:06 Loqi KevinMarks meant to say: well, here you'll hear a lot about microformats
# 04:07 wagle ok.. it was micro-something-or-other... 8/
# 04:09 KevinMarks if you already know html and a bit of css, microformats re easy to understand
# 04:09 wagle not arguing, just trying to make sense of the big picture
nitot joined the channel
# 04:09 KevinMarks they're a set of conventions for using classes to make the page easier to extract data from
# 04:10 KevinMarks microdata instead takes the approach of adding a lot of extra properties that often duplicate the data
# 04:12 wagle i'm just reporting a small selection of what the mooc points me at
# 04:12 wagle .. trying to figure out why this guy seems to be claiming that microdata is "real"
# 04:15 KevinMarks microdata was Ian's attempt to make up a syntax that was somewhere between RDF and mciroformats
# 04:16 KevinMarks unlike the rest of his editorial work, he was less focused on implementations
# 04:16 todrobbins of encoding a set of RDF modelled data for consumption by the client/spider
# 04:17 todrobbins especially when compared to previous encodings of RDF. Microformats and microdata are great too
# 04:19 KevinMarks yes, it's a good dynamic language represention of RDF concepts
# 04:21 todrobbins you don’t think JSON-LD should be included in a script tag on a page? A kind of sidecar metadata file?
# 04:22 Loqi The sidefile-antipattern is a violation of the DRY principle by the use of secondary files (typically in some one-off XML format) to provide information that is a duplicate of information available in primary files on a website (in HTML), and is an antipattern due to typical DRY violation problems such as out-of-date, missing, corrupted, or outright false data https://indiewebcamp.com/sidefile
# 04:22 KevinMarks it's nto really showing mciroformats best practices, as there's lots of that here, but it is a mapling between them
klickreflex joined the channel
# 04:28 KevinMarks if you're going to include JSON-LD in a page, it would make sence to put it in a script tag and assign it to a variable, so it is at least accessible to code running in the page
# 04:29 todrobbins That makes sense. Sidefile is probably the wrong way to describe my example. Loqi, good reference to js;dr
# 04:29 KevinMarks the problem wiht sidefiles, especially ones that have nothing to do with the construction of the page, is that they rot
# 04:29 wagle thsnk for the info! (i'm busy with this test now)
# 04:29 todrobbins Which is why the argument is to just embed the data in the markup itself
# 04:29 aaronpk i guess sidefile doesn't quite describe JSON-LD islands, but it's still DRY
# 04:30 todrobbins aaronpk: by JSON-LD islands, you mean a script tag with a JSON-LD object at the end of an HTML doc?
# 04:30 aaronpk yeah, that's a common term for it as far as i can tell
# 04:31 KevinMarks you cna get any of his pages as json because he will parse the html for you
# 04:32 todrobbins so does microformats support linked data in the attribute values?
# 04:32 aaronpk heh from that same page "The challenge is keeping the JSON-LD data in sync with what appears on the page"
# 04:34 aaronpk mf2 is ultimately a relatively small set of parsing rules, and the parsing rules are not connected with the vocabulary at all
# 04:34 KevinMarks in general, instead of attributes it uses text, html, urls, dates from the page
# 04:37 todrobbins does microformats place vocabulary in a secondary rank? I need to refresh my understanding of the current state of microformats, but it seems there wasn’t much crosswalking/sensemaking between the what hcard might describe and what some other ontology/vocab/what-have-you similarly described
# 04:37 todrobbins I should note I come from a information science background, so you’ll have to forgive my angle at times
# 04:38 KevinMarks we have backwards compatible parsing form microformats 1 in the parsers too
# 04:39 KevinMarks but you cna express any vocabulary in mf2, which was the point of my schema post
# 04:39 Loqi KevinMarks meant to say: but you can express any vocabulary in mf2, which was the point of my schema post
# 04:40 KevinMarks and we have a lot of existing parsers in multiple languages that parse consistently
# 04:43 todrobbins do they display customized search results based on h-card, for instance?
KartikPrabhu joined the channel
# 04:44 KevinMarks it's not very clear tbh; what search engines index is always a bit hit and miss as a lot of it is heuristics
# 04:45 aaronpk I tried adpting the radio episode example from KevinMarks' post to a slightly different JSON-LD representation in mf2
# 04:47 aaronpk (I expanded the @context from the JSON-LD example)
# 04:52 aaronpk "19:45 on the 2nd of June" is really not an ISO date right?
# 04:53 KevinMarks it's certainly not the one true format; it may be one of the innumerable variants, but it doen't look like any of them either
nitot and todrobbins joined the channel
# 05:23 aaronpk i think i got comment/like/repost handling done for my new site
cweiske joined the channel
# 05:26 aaronpk i'm just trying to get back to par with my current site
# 05:27 aaronpk i want to launch this ASAP, *then* I will start adding new features
# 05:27 aaronpk I just don't want to launch with too many features lost
# 05:35 aaronpk getting there. hope to make more progress during tomorrow's HWC
nitot joined the channel
# 06:04 kylewm todrobbins: that's a weird shim to represent a schema.org vocabulary in microformats
# 06:08 KevinMarks hm, so if i subscribe to a wordpress site in woodwindas hfeed, it doesn't get timestamps?
# 06:08 todrobbins good night all! And thank you KevinMarks and aaronpk for being patient with my questions
tantek joined the channel
# 06:09 kylewm KevinMarks: does the wordpress hfeed specify 'published' dates?
# 06:12 tantek aaronpk that's a bummer about etherpad.mozilla.org - odd that it was handled that way with the old.etherpad-mozilla
# 06:12 tantek and I have no idea how many open tabs I have for etherpad.mozilla.org on how many machines. sigh.
# 06:12 tantek not sure we've actually archived all sessions to the wiki
loic_m joined the channel
# 06:24 tantek data islands are defunct technique for adding DRY-violating invisible data to HTML pages (AKA sidecar metadata), first introduced then abandoned by Microsoft as XML Data Islands, and more recently, by Google as JSON(LD) embedded in HTML.
lukebrooker and fkooman joined the channel
friedcell, sanduhrs, eschnou, tvn, lewisnyman, j12t, petermolnar, elf-pavlik and nitot joined the channel
lewisnyman, friedcell, nitot and eschnou joined the channel
lewisnyman_ and j12t joined the channel
KevinMarks__, Tino, alexhartley, j12t, KevinMarks___ and KevinMarks2 joined the channel
ttepasse and j12t joined the channel
loic_m, KevinMarks__, mlncn, j12t, KevinMarks, KevinMarks_, nitot, friedcell, davidmead and eschnou joined the channel
# 13:07 ben_thatmustbeme !tell aaronpk do you have a list of all the push notifications (phone notifications) that you have needed? Such as 'tagged you in a photo' ' mentioned you' etc
# 13:07 Loqi Ok, I'll tell them that when I see them next
elf-pavlik, hs0ucy, snarfed, snarfed1, friedcell, tvn, mlncn, peacekeep3r and lewisnyman joined the channel
# 14:43 aaronpk so I saw this on twitter this morning and thought it was a joke/parody
# 14:43 aaronpk "The Accelerated Mobile Pages (AMP) Project is an open source initiative that embodies the vision that publishers can create mobile optimized content once and have it load instantly everywhere."
# 14:43 aaronpk I thought it was going to be a joke promotion to get publishers to write plain HTML pages with no javascript bogging down the page
# 14:44 aaronpk "AMP HTML is a new way to make web pages that are optimized to load instantly on users’ mobile devices."
elf-pavlik joined the channel
# 14:47 Zegnat But with custom elements / web components as well, aaronpk
snarfed joined the channel
# 14:49 Zegnat If I understand it right, they forbid the use of JS outright? Fits perfectly with tantek's js;dr then. No JS apart from components they themselves have created, that is.
# 14:51 aaronpk if you look at the demo, they launch an in-browser full screen popup that shows the contents of the story right on the google search results
# 14:52 Zegnat There is some weird stuff in there too. They limit what CSS you are allowed to use (for speed? don’t know) but allow any @font-face inclusion. Which is a much bigger bottleneck than any CSS
# 14:52 aaronpk basically they want the structured content from the publisher so that they can reformat it to show in their own page
lewisnyman joined the channel
# 15:04 Zegnat Well, I don’t really care who starts the initiative and for what reasons. Less crap around articles will be a plus in my book
# 15:05 aaronpk we'll see if it translates to their main site too, cause they may only be updating the HTML on their mobile site
# 15:06 Zegnat I think there is a big incentive for publishers to *not* have it as their main site yet. The fact AMP limits advertisement and tracking possibilities comes to mind.
# 15:07 Loqi Zegnat meant to say: I think there is a big incentive for publishers to *not* have it as their main site yet. The fact AMP limits advertisement and analytics possibilities comes to mind.
# 15:08 ShaneHudson_ My image wasn't appearing, changed to HTTP instead of HTTPS and it worked. It should be HTTPS
# 15:09 aaronpk the wiki doesn't do anything with hotlinked images, it just writes it to the <img> tag
# 15:10 aaronpk (i'm going to have to proxy those eventually so that the wiki is https-only but that's a project for another day)
# 15:10 ShaneHudson_ Zegnat: So is it something I'm doing nginx side? I see quite a few broken images on that page
# 15:11 aaronpk well one of the broken images is because someone has an invalid ssl cert
# 15:11 aaronpk another one looks like they lost their hosting account
# 15:12 aaronpk yup all legit errors or no images in the first place
# 15:12 aaronpk i should totally use my avatar archive that webmention.io is using to save the external images that are linked from the wiki
todrobbins joined the channel
# 15:14 ShaneHudson_ I've always liked the idea of somehow proxying dead links to archive.org or alternative archived sites. But think it could very quickly become a massive project
# 15:14 Zegnat I wonder why your image is being blocked, ShaneHudson_, I can’t really see any differences between the HTTP and HTTPS one
# 15:14 Zegnat Although Safari’s developer tools have problems with it too
# 15:15 todrobbins aaronpk: I just saw the comment by tantek about etherpad.mozilla.org. what happened there?
# 15:16 aaronpk they're archiving their old etherpad installation
lewisnyman joined the channel
# 15:18 ShaneHudson_ Well it is called ether pad, at least they made no false pretenses. Unlike Google Keep
tvn_ joined the channel
# 15:20 Zegnat I know, and the headers on the right seem to be exactly the same for both, so I am clueless as to a difference
# 15:20 ShaneHudson_ Yeah. I am going to ignore it for now. Sounds like it could be a cloudflare thing
# 15:20 ShaneHudson_ Used to host DNS myself but I couldn't keep up with blocking the attacks
LanceyWork joined the channel
# 15:24 todrobbins aaronpk: does the chat from the public pads get collected by indiewebcamp as well?
# 15:24 aaronpk there also hasn't been much on those chats since we have IRC
wolftune, ttepasse and j12t joined the channel
lewisnyman joined the channel
# 16:01 KevinMarks apparently they allow some svg now but didn't update the spec o_O
gRegorLove joined the channel
# 16:01 Zegnat I thought aaronpk had a tweet saying they don’t *require* schema?
# 16:02 Zegnat They ban a lot of stuff though, and some things feel a bit random to me (CSS selectors? Seriously?) while other things that are performance hit on mobile are completely fine (@font-face)
snarfed joined the channel
# 16:05 KevinMarks "AMP HTML documents MUST be marked up with schema.org/CreativeWork or any of its more specific types such as schema.org/NewsArticle or schema.org/BlogPosting. More types may be added in the future."
# 16:05 KevinMarks Both JSON-LD and microdata serializations of schema.org are supported.
# 16:07 Zegnat Ah, alrighty, got your tweet mixed up with him reply then!
# 16:07 Loqi Zegnat meant to say: Ah, alrighty, got your tweet mixed up with his reply then!
snarfed joined the channel
shiflett joined the channel
# 16:12 KevinMarks so I'm on TWiG to argue with Jarvis about this, so lets work through it
# 16:13 KevinMarks jarvis: "Imagine starting a new media service without a web site"
# 16:17 bear that buzzmachine link is taking a lot of time to load...
tantek joined the channel
elf-pavlik joined the channel
# 16:29 aaronpk augh the new namecheap control panel is a huge step backwards
# 16:30 aaronpk i used to be able to see my whole list of domains on one screen, now I can only see 5 and I have to scroll
# 16:36 aaronpk uhoh i wonder if my SSL change just now broke something
# 16:37 aaronpk yup. Loqi is running on a server that's too old and can no longer talk to the wiki
snarfed joined the channel
# 16:49 aaronpk in today's lesson in domain registration management, Aaron learns that when you transfer a domain between registrars, the contact info stays the same. If your previous registrar used "whois privacy" and published their own proxy email, the proxy email transfers over too.
# 16:50 aaronpk this means when your new registrar tries to contact you via the email address onthe domain, the emails don't go through, because the previous registrar stops forwarding the emails since the domain isn't with them anymore
nitot joined the channel
# 16:56 tantek I fully expect that eventually bear will end up educating all of us (and Loqi) on sysadmin / devops 101
# 16:56 tantek and maybe, just maybe we can help simplify all of that for the next generation
[kevinmarks] joined the channel
# 17:00 bear one of my longer term goals at &yet is to make open source ops more of a real thing
# 17:00 bear breaking down what I do daily into a series of recipes that people can learn from and use
# 17:01 bear an opinionated collection of best practices
cleverdevil joined the channel
# 17:05 ShaneHudson_ I always forget you are at &yet. One of my favourite companies on earth heh
lewisnyman joined the channel
# 17:22 aaronpk whoa the json-ld example has a ton of invisible metadata
# 17:22 aaronpk the microdata example they moved it into the body of the page
# 17:24 aaronpk hm does h-entry have any concept of "publisher" (different from author)
# 17:27 aaronpk ben_thatmustbeme: not required, you can use AMP instead
# 17:27 aaronpk tantek: that doens't work in this case. this is an article which is specifying the publisher of the same article
# 17:27 aaronpk but it's the canonical URL of the article, not a citation of it
# 17:28 tantek aaronpk - I'm saying there have been use-cases to justify a "publisher" in h-cite, but not yet in h-entry
# 17:32 aaronpk oh boy, it's the weird style tag that's breaking it
# 17:33 aaronpk when this is in the <head> it doesn't find the h-entry tag
# 17:33 aaronpk <style>body {opacity: 0}</style><noscript><style>body {opacity: 1}
</style></noscript>
# 17:34 aaronpk i don't even understand why that would be a thing
# 17:35 tantek so that the page can use script to fade the body opacity from 0 to 1
# 17:36 tantek having thoroughly code reviewed it and its test suite / harness
# 17:37 aaronpk i'm just gonna run the node parser on an AWS Lambda URL and use that for everyhting :P
# 17:39 [kevinmarks] We have 3 parser options, it may break the shitty built-in one, but lxml and html5lib should work
# 17:39 aaronpk the fact that both php and python break would indicate it's an mf2 problem rather than an html problem
# 17:39 tantek nah - more likely they copied the same parser bug
# 17:39 aaronpk the top-level object on the page is the nested author h-card
# 17:40 tantek if it's not seeing the body class=h-entry, then it's the parser's fault
# 17:41 kylewm html5lib and lxml handle <noscript> very differently
# 17:42 aaronpk well I don't know how the parsers work anyway. where should I file bugs?
# 17:42 aaronpk we need to get all the parsers using the same test suite so we can start adding these examples to all of them easily
# 17:44 aaronpk <link class="u-url" href="..."> should work too right?
# 17:45 [kevinmarks] I spent a chunk of time on getting the python parser working with the general test suite
hs0ucy joined the channel
# 17:48 kylewm oh bummer, looks like testrunner-47055.onmodulus.net is gone/down
KartikPrabhu joined the channel
# 17:49 aaronpk wow the googlebot github account makes you sign a CLA when you submit a PR
# 17:49 aaronpk super automated process, but the bot is commenting on github like an actual account!
# 17:50 kylewm rustlang has a bot that automatically randomly assigns a reviewer for PRs
nitot joined the channel
# 17:55 tantek indeed. one does not need custom elements for mobile performance. that smells of an excuse to use "new shiny".
# 18:02 aaronpk anyone want to review that example before I send the PR?
KartikPrabhu and aaronpk joined the channel
aaronpk joined the channel
# 18:21 KevinMarks it lets you embed tweets. it overrides benward's nice blockquote fallback
# 18:23 tantek does the ampproject site itself use amp? it's too slow to load for me to bother finding out.
# 18:26 snarfed googlebot++ aaronpk, agreed, the new automated CLA process is light years better than the old process(es)
# 18:26 aaronpk okay I *think* I updated my SSL cert everywhere it's used. Guess I'll find out on Friday if things start breaking
# 18:28 bear at least you updated it before the ocsp window hit
# 18:28 bear that time when ocsp caches have your old cert but you've just updated to a new one
# 18:29 bear startssl is very fast with updates - so you won't get a lot of breakage today
# 18:32 aaronpk i am, but it's all very specific to my servers, not really useful generally
# 18:34 aaronpk the scary thing is this cert expires in 2 years which means I am very likely going to not remember this
hs0ucy joined the channel
# 18:34 bear OCSP window is that small amount of time between when your SSL certificate vendor registers your new cert and when the browser update their certificate cache
# 18:34 tantek HTTPS - how to make your website more fragile without anyone attacking it!
# 18:35 tantek HTTPS is like the inkjet business model for CAs.
# 18:35 bear pfft - I could say the same thing about most developers ;)
# 18:35 aaronpk i'm using this cert not only on web stuff tho. mail server, IRC server, IRC bouncer
# 18:36 tantek aaronpk, supposedly LetsEncrypt is going to solve the cert update problem
# 18:36 tantek I'll believe it when it ships and does its first automated update
XgF joined the channel
# 18:38 tantek they want to host an HWC or IWC one day if possible
# 18:38 tantek one of their folks (Lauren B) has been to a bunch of our HWC SF meetups
# 18:39 aaronpk man i wonder what it'd take so that I can stop running this mail server
# 18:39 tantek aside, just told CSSWG folks who lost their Moz Etherpad tool they could use ours if they don't mind that everything they type into it goes CC0 / Public Domain
# 18:39 bear cloudflare works because they have automated the hard part of doing HTTPS over a CDN while reducing the amount of cdn grief you normally get
# 18:39 tantek aaronpk, consider lurking in # css on W3C IRC
# 18:39 tantek (e.g. when you're connected to # social on there)
# 18:41 cleverdevil well, honestly, with anything infrastructure related for IndieWeb efforts.
# 18:41 cleverdevil I can't guarantee help in every case, but I can always be an advocate :)
# 18:49 GWG !tell tantek I have a use case for publisher in an h-entry or h-feed, see site name.
# 18:49 Loqi Ok, I'll tell him that when I see him next
_6a68_ joined the channel
nitot joined the channel
# 18:55 voxpelli apart from a few "op-" properties like "op-published" and "op-ad" it's mostly just basic html tags and thus fairly similar to how one could interpret an h-entry
KartikPrabhu and eschnou joined the channel
# 19:08 aaronpk I just re-read the PuSH spec and it says the notifications need to have HTTP headers for rel=hub and rel=self. pretty sure Switchboard is not sending those.
# 19:08 aaronpk kylewm: have you noticed any issues receiving pings from Switchboard?
# 19:09 aaronpk you must be using a unique subscription URL per feed then
# 19:09 aaronpk and you aren't using any contents of the POST from the hubs other than the URL that was POSTed to, right?
# 19:10 voxpelli including the rel-hub must be to enable the move to a new hub and the rel-self to know what content that is actually being pushed?
# 19:11 kylewm aaronpk: yeah unique subscription url -- I use the X-Hub-Signature if provided but no other headers
# 19:11 aaronpk yeah if you're using a unique subscription URL that's enough to identify what is being updated. if you're using the same one you'd need to use the rel=self value from the header
# 19:12 aaronpk funny cause the spec also says it's "good practice" to use a unique subscription URL
# 19:13 voxpelli talked to someone about paginated pages and PuSH and the pushed URL doesn't necessairly have to match the subscribed one
# 19:13 voxpelli you can subscribe to the first page, but receive an update for the third page for example
# 19:14 voxpelli hmm, that doesn't matter here when I think again because the rel-self would still be the same I guess
# 19:14 aaronpk and still if you're using a unique subscription URL it doesn't matter
# 19:14 voxpelli yeah, it's probably just to make simple implementations easier
# 19:15 voxpelli the rel-hub is the more interesting one – enabling the move from one hub to another
# 19:15 kylewm voxpelli: what would a push for the third page look like?
# 19:15 aaronpk voxpelli: is the implication that if the subscriber gets a rel=hub that's a different value than was used, they should create a subscription on the new hub?
# 19:16 voxpelli kylewm: just the content of the third page with the same rel-self I think, was a while since I discussed it
# 19:16 aaronpk huh, that's not mentioned in the spec at all, which just says "with rel=hub pointing to the Hub" which kind of implies it would be the same hub that was subscribed to
# 19:18 voxpelli maybe it's not meant to communicate a hub switch, but it would make sense to use it for one and it would be useful to have a way to communicate something like that :)
# 19:19 aaronpk it would probably be useful to be able to indicate a hub switch. otoh as soon as the subscription expires the subscriber is going to find the new hub anyway
# 19:19 voxpelli that reasoning also shines some light on why rel-self isn't the same as rel-canonical
# 19:19 kylewm voxpelli: i think i would interpret that exchange differently
# 19:20 kylewm you said "i expected push subscriptions to be for the exact url" and he said "yes exactly"
# 19:21 voxpelli right, I realize there's some ambiguity in that exchange now, I guess I have to sort that out with Julien
# 19:22 voxpelli one challenge with PuSH when it comes to Crud is what happens with items that has fallen of the first page of a feed
# 19:24 kylewm very similar to the problem i'm having thinking through how to fetch likes and comments and show them in woodwind
shiflett and wolftune joined the channel
# 19:42 snarfed it'll be an awesome feature, but i definitely don't envy you implementing it kylewm
# 19:44 voxpelli kylewm: in the spirit of Salmention a received like or comment should trigger a PuSH
# 19:45 voxpelli so yeah, very similar, if PuSH has support for pages (which it might already perhaps have with a liberal interpretation of the spec) then that would be enough
# 19:46 voxpelli snarfed: only in classic feed readers I guess? which weren't checking for unique id:s or URL:s to find duplicates?
# 19:46 aaronpk the spec is meant to be interpreted liberally so go for it?
# 19:46 kylewm snarfed: ha KevinMarks mentioned that. i probably won't do an ambitious implementation, at least at first. tantek suggested just fetching like and comment count on demand (maybe on a click). that would be a good proof of concept
renoirb, martinBrown, tantek, nitot and Lancey joined the channel
# 19:58 tantek GWG, that's publication - not publisher. which we have in h-cite - and likely would work for site name too
# 19:58 voxpelli Instant articles is almost a microformats approach – only weird thing is their geo-data
# 19:59 tantek subsequently I asked do we have history of proprietary silo markup failures?
# 20:01 snarfed voxpelli: even in modern readers with id handling. meh!
# 20:02 tantek voxpelli: anything geo related in RSS has a murky past, including the poorly named and defined GeoRSS
# 20:02 tantek (back when it was fashionable to suffix your "thing" with *RSS to make it sound cool, like MediaRSS or whatever it was called that Yahoo did)
# 20:02 voxpelli tantek: Instant Articles aren't RSS-based, RSS is simply a distribution mechanism for the HTML
# 20:08 voxpelli good thing is that that's the only weird thing – the rest is class-based and HTML5-tags I thing
# 20:09 voxpelli they link to the GeoJSON-site rather than documenting it themselves
# 20:10 tantek "element with the op-map class attached" ... "with the op-geotag class attached."
# 20:10 tantek voxpelli: is that what you meant by microformats approach?
# 20:10 ShaneHudson_ GeoJSON is a standard format for geographic data, ranging from single points to collections of complicated polygons.
# 20:11 voxpelli tantek: the "op-published" and "op-modified" classes are very similar to their "dt-" counterparts
# 20:11 voxpelli and they use "op-" for some "h-" style classes as well
# 20:11 tantek voxpelli: that means you have to know what the fixed op-* vocabulary is in order to parse it
# 20:12 voxpelli yeah, but thinking it would be fairly easy to translate the "op-" classes to mf2-classes as they both share some commonalities
# 20:14 voxpelli it's interesting to note that there's two prefixes just like in "open graph" – "op-" and "fb-"
# 20:14 voxpelli so they do distinguish between reusable prefixes and such that are purely meant for Facebook
# 20:15 Loqi voxpelli meant to say: so they do distinguish between reusable classes and such that are purely meant for Facebook
# 20:18 KevinMarks yuck contain a <meta name="viewport" content="width=device-width,initial-scale=1,minimum-scale=1,maximum-scale=1,user-scalable=no,minimal-ui">
# 20:26 aaronpk at least now it'll be on record if they reject it
tantek joined the channel
# 20:38 tantek oh hey at least AMP appears to be listening to the js;dr lessons! (finally reading IRC scrollback)
# 20:44 ShaneHudson_ I don't know much about amp but currently working with a CMS based on Wordpress that bans loads of things... driving me insane. Everything I do is ending up as a total hack.
# 20:44 aaronpk ah there is a whitelist of supported fonts for @font-face
# 20:45 aaronpk "Font providers can be whitelisted if they support CSS-only integrations and serve over HTTPS. The following origins are currently allowed for font serving via link tags: https://fonts.googleapis.com"
# 20:45 aaronpk wait but then "Authors are free to include all custom fonts via a @font-face CSS instruction via their custom CSS. Fonts included via @font-face must be fetched via HTTP or HTTPS scheme."
# 20:47 aaronpk i don't actually read those as headings, more like strangely large callouts
# 20:47 tantek except they're not callouts because they're not elsewhere in the doc
# 20:48 tantek AND when they do, they function more as headings than not, e.g. AMP HTML loves CSS!
# 20:50 bear he (jeff) was at the event announcing this
# 20:50 tantek KevinMarks, keep saying js;dr as often as you can ;)
nitot joined the channel
KevinMarks__ joined the channel
# 21:02 bear KevinMarks - so will DRM'd video streams work with this google blessed js video tag?
# 21:06 aaronpk KartikPrabhu: it makes sense when you do the demo on a phone
# 21:06 aaronpk or on desktop change your user agent to something that looks mobile
# 21:07 KartikPrabhu so if it speeds up things on mobile why not speed up things on desktop too?
# 21:08 cleverdevil looks like its basically trying to be a more open version of FB instant articles and Apple News and the like.
# 21:09 aaronpk well they support separate AMP files from the main site
# 21:09 aaronpk so nytimes and such are publishing two pages for every article
# 21:11 kylewm do you need to be using chrome on a mobile device?
# 21:11 aaronpk it actually worked pretty crappy in chrome on ios
# 21:11 aaronpk KartikPrabhu: you should see special results at the top for the AMP-enabled search results
# 21:13 bear only if your using nodejs to generate server side static pages
# 21:13 bear if your using dynamic pages at all it won't load your js
# 21:14 aaronpk right. at least they're encouraging people to publish html
# 21:14 KartikPrabhu seems like "lets make bloated webpages and add Google's bloat to unbloat them"
# 21:16 aaronpk it's so they can have more control over the loading of the images and videos when displaying a page
# 21:18 tantek Apple News Format is an yet unspecified [[Apple ]]-proprietary format for publishing news for iOS devices.
tantek joined the channel
# 21:25 bear makes me wonder if google has thought about doing micropayments if AMP is used
gRegorLove joined the channel
# 21:29 bear that's the point I was badly making - if AMP is used then it allows finer grained tracking of click/views with google blessed tech
# 21:32 bear hmm, google via AMP as a article broker
# 21:45 KartikPrabhu Have been thinking about it since I got a bunch of addressed spoofed emails from the Nobel Prize committe ;)
# 21:46 aaronpk you can check that the domain of the source URL matches the domain of the author
# 21:46 aaronpk that would block bridgy but would also block spoofing
# 21:47 aaronpk you could then whitelist bridgy and allow only bridgy (or whatever backfeed service your'e using) to spoof
# 21:47 snarfed KartikPrabhu: if you want something bigger, e.g. check that the author is truly the person with the name and picture in the h-card, that's kinda either unsolvable or massively hard
# 21:47 snarfed you basically need to check id out of band, which...no
# 21:47 aaronpk haha yeah. i just went through this with startssl extended validation
# 21:48 KartikPrabhu not a problem with webmentions yet, but was thinking in comparison to email
# 21:48 snarfed although, hmm. using the "silos as plumbing" model, you could check that their profile is "verified" (or whatever it's called)
# 21:48 KartikPrabhu yes it is. I was wondering if webmention can be better than email there
# 21:48 aaronpk attempts to solve email are actually worse than the source URL + author URL check IMO
# 21:49 aaronpk you don't need vouch to reject spoofed webmentions
# 21:49 snarfed aaronpk: you do if they serve the h-card url, image, etc all from their domain
# 21:49 snarfed having said that, this is so far not a problem in practice
# 21:50 tantek pretty sure this is all already documented on the wiki - did any of you bother to check?
# 21:52 snarfed otherwise: tons of prior art, huge nerd snipe tarpit, and not an actual problem we have yet...all good reasons to deprioritize with a vengeance
# 21:52 aaronpk the domain check idea only works in conjunction with receiving a webmention
# 21:53 tantek OTOH I'm more than happy to fuel the nerd snipe fire for minimizing HTML that AMP appears to be sparking
# 21:53 tantek we've already been doing that in IndieWeb for quite some time (minimizing HTML) out of laziness / efficiency. but I repeat myself.
# 21:54 KartikPrabhu but also no potential solution documented but that is ok as low priority
# 21:54 tantek KartikPrabhu: it's ok, feel free to add anything new you think of (whether question, solution, ideas) to that page
nitot joined the channel
# 22:08 KevinMarks so, do we have a thing to recommend to leo et al that will actually work?
# 22:10 tantek KevinMarks by "actually work" presumably you mean "has been user tested on someone other than Leo"?
# 22:10 tantek you could user test with someone else, whatever you were thinking of demoing to Leo
# 22:12 tantek KevinMarks: have you checked the last week in the indieweb page(s)?
# 22:13 aaronpk that will work as long as you put the quill email address in the "To" field
# 22:15 tantek comments are every publishers & producers problem
# 22:15 tantek keep repeating the message that we have and are using a solution, and your message may eventually get through
friedcell and lukebrooker joined the channel
Nissyen and mlncn joined the channel
snarfed joined the channel
yalamerde and nitot joined the channel
# 23:01 tantek it's fine. first half of the year, and 2nd half of the year!
KevinMarks joined the channel
# 23:06 KartikPrabhu eventually we'll hit a singularity and the wiki will become sentient enough to plan its own planning
# 23:06 Nissyen Discussion of that should be on the planningPlanningSentience page
# 23:09 aaronpk that way there's no confusion between planning a specific event vs a bunch of events
KevinMarks_ joined the channel
# 23:14 tantek the point was that for any IWC related home page blahblahYearLocation
# 23:14 tantek that all subpages be blahblahYearLocation/subpage
# 23:15 aaronpk i'm sorry I brought this up. I just went to go look up the status of 2015/SF and couldn't find it because I was on the wrong planning page
snarfed1, snarfed, KevinMarks and KevinMarks_ joined the channel
# 23:51 KevinMarks so <amp-twitter></amp-twitter> doesn't treat it's contents as fallback
# 23:55 KartikPrabhu that seems to be a probelm with web components in general. Custom tags but no need to specify good default fallback behaviour
nitot joined the channel