2017-07-26 UTC
[miklb], deep-book-gk_ and tantek joined the channel
# 00:50 gRegorLove Interesting that allinthehead.com article didn't pick up my author photo
# 00:51 gRegorLove Looks like most indie responses it didn't get photos for actually
# 01:10 tantek gRegorLove: probably an indication of an implementation of a subset of the authorship algorithm
[kevinmarks] joined the channel
# 01:10 tantek makes me wonder if it's time for a test suite for that
# 01:11 tantek and a validator that follows /authorship fully for content (and indicates which step your post used)
tantek, j12t_, deathrow1, eli_oat, KevinMarks, snarfed and [miklb] joined the channel
# 05:15 Loqi Ok, I'll tell them that when I see them next
gRegorLove joined the channel
# 05:17 Ruxton miklb: what are you sending to the micropub endpoint?
# 05:18 Ruxton interesting, I wonder if Quill is sending filename in the post
# 05:18 Ruxton are you able to upload same photo inside the wordpress admin UI?
# 05:19 [miklb] good question. I haven’t tried posting directly from my phone to the admin.
# 05:20 Ruxton try that, if it works, it's likely something Quill is not doing right, as I'm throwing images at wordpress-micropub every few days without drama
# 05:21 [miklb] yes, I was able to upload that photo through wp-admin from iOS fine. OwnYourGram works fine
# 05:22 gRegorLove That error messgage is from WP core. May only be doing file extension checks. I ran into that with my MP for ProcessWire plugin.
# 05:23 [miklb] it is a core error and there’s a whitelist of extensions allowed iirc.
# 05:23 [miklb] wonder if there’s something I can filter to allow coming from Quill…
# 05:24 Ruxton it's easier to fix it in Quill and make Quill behave correctly (send original file extension)
cweiske joined the channel
# 05:30 [miklb] possibly, I tried with the plugin in that issue but got same error
# 05:33 gRegorLove What mime type does finfo_file() show for the image on your server?
# 05:37 [miklb] I am using a plugin to offload images to my S3 bucket, but I don’t think that’s teh problem since my OYG images work fine.
# 05:39 Ruxton i've attempted a fix into Quill for this to ensure it sends the original filename instead of the tmp_name
# 05:41 Loqi [tjgillies] #81 quill may or may not be converting animated gif to static gif
Defenestrate joined the channel
# 05:47 [miklb] as an aside, I didn’t realize the native wp-admin was so functional on my phone. Might not need Quill on iOS afterall
# 05:49 Loqi [Ruxton] #82 Ensure original filename+extension is sent
# 05:51 Loqi ruxton has 1 karma in this channel (4 overall)
tantek and KevinMarks joined the channel
# 07:19 Zegnat !tell tantek if you are looking for a “validator” for authorship, I like testing with https://sturdy-backbone.glitch.me/ which will highlight the exact path through the algorithm that needs to be taken for any page.
# 07:19 Loqi Ok, I'll tell them that when I see them next
jjuran, KartikPrabhu, Defenestrate, drewinthehead and sebboh` joined the channel
AngeloGladding, dansup and drewinthehead joined the channel
# 09:36 drewinthehead gRegorLove: I basically haven't implemented the authorship algo yet. Just looking at that now.
KevinMarks joined the channel
# 09:57 drewinthehead I think I'm right in staying there's no a standalone PHP implementation for the authorship algo, just a few partial in-the-wilds. Is that right? Seems like it would make a useful addition.
KevinMarks_ and KevinMarks joined the channel
KevinMarks and jeremycherfas joined the channel
barpthewire joined the channel
# 12:42 jeremycherfas I think I’ll just live dangerosly and see if I can get things going on the live site.
# 12:45 jeremycherfas No worry, thanks. I just added the live site domain and will work with that.
# 12:48 jeremycherfas Ah well. I’ve got an hour or so to play live. If I can get something — anything — I’ll be very happy.
# 12:53 voxpelli And what are you trying to add? Just the "39e1e221.ngrok.io", right?
# 12:54 jeremycherfas Ah no. That won’t work, because all my local sites are under a subdomain here, because of MAMP
# 12:57 sknebel that shouldn't be a problem though, since *you* choose for which pages on the domain it's the webmention endpoint?
# 13:02 sknebel webmentions only get sent to webmention.herokuapp for pages on which there is a header declaring it as the endpoint
dougbeal|mb1 joined the channel
# 13:18 jeremycherfas Whoa! Truly wonderful stuff voxpelli. I used the sample data first to see how much styling I would have to do, and they all came throough just perfectly.
# 13:19 Loqi voxpelli has 9 karma in this channel (106 overall)
KevinMarks joined the channel
KevinMarks joined the channel
# 13:32 voxpelli jeremycherfas: well, with CSS Flexbox and such it sure is ;)
# 13:33 jeremycherfas I guess that as the random subdomain that ngrok supplies will change next time I want to use it that I will have to start from scratch, logging out this site
KevinMarks, KevinMarks_ and eli_oat joined the channel
# 13:42 voxpelli that + CSS Grid which is soon available to wide use as well will make layouts in CSS so much more powerful and easy
KevinMarks joined the channel
# 13:48 jeremycherfas I’ve read about CSS Grid, and it seems remarkably powerful. Flexbox might be easier to start with.
# 13:50 jeremycherfas Oh, I see there’s a tool for helping it cope with older browsers. That’s handy.
# 13:50 sknebel you can feature-check for it, and AFAIK it overrides other layouting rules - if it isn't supported, you can fallback to a conventional layout
eli_oat1, KevinMarks and dougbeal|iOS joined the channel
singpolyma, cweiske and [asteres] joined the channel
# 14:04 voxpelli Flexbox is supported for long enough for me to consider it to be okay to expect everyone to have it
# 14:05 voxpelli Especially as it became introduced before @supports was introduced
KevinMarks, KartikPrabhu and j12t joined the channel
# 14:36 Loqi sknebel, did you find your MP content type notes?
# 14:36 Loqi Countdown set by sknebel on 2017-07-25 at 4:35pm CEST
dougbeal|mb1, jjuran and KevinMarks joined the channel
KevinMarks, j12t and tantek joined the channel
KevinMarks joined the channel
# 16:06 Loqi tantek: Zegnat left you a message 8 hours, 46 minutes ago: if you are looking for a “validator” for authorship, I like testing with https://sturdy-backbone.glitch.me/ which will highlight the exact path through the algorithm that needs to be taken for any page.
# 16:06 Loqi PASTA is an acronym/abbreviation for Publish Anywhere, Save To (private) Archive, the practice of automatically saving a copy of whatever you post on (social media) silos to someplace else under your own control, like a private directory on your own server, or a local folder on your laptop that is less vulnerable to site-death https://indieweb.org/PASTA
snarfed joined the channel
KevinMarks joined the channel
# 16:21 Loqi perch has 1 karma in this channel (2 overall)
KevinMarks joined the channel
# 16:23 drewinthehead Still need to improve author detection, but that's coming. Small steps.
j12t joined the channel
# 16:30 drewinthehead @tantek Looks good to me, appreciate it being well spec'd, just need to get it implemented.
# 16:30 tantek cool. good to know! appreciate any feedback on any difficult / awkward / confusing bits.
snarfed, KevinMarks, [miklb] and gRegorLove joined the channel
KevinMarks joined the channel
cweiske and j12t joined the channel
# 17:32 tantek.com edited /Tumblr (-65) "/* Criticisms */ merged with prior Releasing inactive accounts, move that section near involuntary identity reassignment due to being related" (
view diff )
j12t joined the channel
# 17:42 voxpelli It's no longer blocked as far as I remember, happened when I worked at Bloglovin
# 17:46 gRegorLove Wow, I don't get the annoying 1/3 overlay to join Facebook when I open that ^
# 17:49 snarfed there's been lots of FUD forever about how many people you leave behind if you're SSL-only, and more recently SNI-only, but very little actual measurement ordata
KevinMarks joined the channel
snarfed joined the channel
# 17:58 Zegnat snarfed, I have noticed that too. When I first set-up SSL I saw all these warnings about SNI, but I haven’t had any complaints and everyone I ever linked to my site has been able to reach it.
# 18:02 jeremycherfas Tantek: your experience of being able to set a new password suggests that if I allow my Tumblr site to be dormant I will eventually get the same email and be allowed to reset my passowrd.
KevinMarks_ joined the channel
# 18:03 tantek assuming you still have access to the email you used to sign-up on Tumblr!
# 18:03 aaronpk the only SNI problems i've had is from people using python2 HTTP clients
# 18:06 cweiske elasticsearch head plugin running on localhost:9200 is not able to access ES instance on localhost:8080 (ssl forwarding) because of CORS :/
KevinMarks, KevinMarks_, snarfed and KartikPrabhu joined the channel
# 18:51 sknebel Regarding authorship tests, there is a github repo with test cases, Zegnat already linked to the "explaining" tool
# 18:57 dgold regarding https 'leaving users behind' - that paper suggests a vanishingly small cohort of users
# 18:58 dgold and I don't think the size of that cohort countermands the benefits of https
snarfed joined the channel
# 19:04 cweiske chat search will continue to not return emojis, because of elasticsearch tokenization problems with unicode and a too-low ES version on my server :/
jeremycherfas joined the channel
# 19:06 Loqi [cweiske] #27 emojis not displayed in results
# 19:09 sknebel copy-paste because lazy: tantek: Zegnat left you a message 8 hours, 46 minutes ago: if you are looking for a “validator” for authorship, I like testing with https://sturdy-backbone.glitch.me/ which will highlight the exact path through the algorithm that needs to be taken for any page.
# 19:09 Loqi [sandeepshetty] authorship-test-cases: Test cases for the IndieWeb Authorship Algorithm
snarfed joined the channel
# 19:15 tantek sknebel: I don't think we've heard from sandeep in a while - probably worth forking somewhere to collaborate
KartikPrabhu joined the channel
# 19:22 Loqi Ok, I'll tell them that when I see them next
# 19:23 aaronpk that seems to be following the authorship algorithm correctly
# 19:23 Zegnat And it needs an h-entry, tantek.com is not an h-entry
# 19:23 aaronpk "start with a particular h-entry to determine authorship for, and no author"
KevinMarks joined the channel
# 19:24 Zegnat Probably because it discovers there is no valid h-entry for the URL at that step in the parser
# 19:25 Loqi sknebel has 18 karma in this channel (50 overall)
# 19:29 sknebel looking at tanteks permalinks reminds me that there still is an open bug in the JS mf2 parsers triggered by them
# 19:34 Zegnat sturdy-backbone.glitch.me just showing the entire JSON for the h-cards confronts me with how much stuff I am pushing to everyone who tried to get my author information, haha
# 19:36 Zegnat Also I might see a big. Is the JS mf2 parser easily testable somewhere? sknebel?
# 19:41 sknebel (not quite stock, my open PRs are already merged in that one)
# 19:42 Zegnat Still doesn’t match the PHP output on weight. Now I have to see which parser is wrong.
# 19:49 Zegnat The rel="author" mentioned in that section is implemented in the algo though. Can’t it be removed?
# 19:49 Loqi tantek has 12 karma in this channel (374 overall)
# 19:49 tantek noting when it was added - should be resolved
# 19:50 sknebel Zegnat: looks like a bug in the JS parser, presumably it doesn't do the .h-x>abbr:only-child rule correctly?
# 19:51 sknebel in general I assume it has quite a few bugs. I tried it to update to the current mf2 test suite but it failed quite a bit of that
# 19:52 sknebel not sure if I updated it wrong or actual bugs in all cases, but
# 19:52 Zegnat I will have a closer look in a bit. It might just as well be a mistake on the PHP end, mind you. I am doing some weird stuff with the mark-up there.
# 19:56 dgold whew - nice even playing with php - I have photos (by url), categories and JSON working on my micropub implementation
# 20:01 Zegnat I am not seeing it getting changed sknebel. Am in June 2015 now and still only-child
# 20:01 sknebel doesn'T seem so. but e.g. photo and name are parsed differently, and the code treats them all the same it seems
# 20:02 Zegnat It used to be called :only-node in the parsing spec, apparently. But I don’t think it has ever been of-type
# 20:03 sknebel (also, I totally wasn't aware that there is an implied *photo* property. so an h-entry with a single img in it might accidentally get a photo property?!)
# 20:05 Zegnat I wonder if that trips up post type discovery
# 20:06 Zegnat Write a blog post, have 1 IMG in there, suddenly parser output has a u-photo and PTD turns it into a photo post?
# 20:06 tantek "an h-entry with a single img in it" sure sounds like a photo post
# 20:07 Zegnat Which is why I said “1”, tantek. It could have 100 paragraph tags and a single IMG, and the h-entry would get a photo property
# 20:07 sknebel Zegnat: well, it can't be nested in e.g. the e-content
# 20:08 Zegnat Could be 2 paragraph note, no e-content at all.
# 20:08 sknebel or could be a decoration image that's part of the layout
# 20:09 sknebel (getting OT, what should they be instead if they are pixel images? or is it "should not be pixel images"?)
[miklb] joined the channel
# 20:10 sknebel like, some design element thats a picture. a fancy "bookmark" symbol on a bookmark post or something like that
# 20:13 sknebel I remember the distinction between u-featured and u-photo being important?
# 20:13 Zegnat IMGs are implied as photo properties if they are only-of-type, not only-child, aaronpk
tantek joined the channel
# 20:14 Zegnat Also a good one sknebel! Should a u-featured suddenly count as an implied u-photo. My first reaction is no
# 20:14 sknebel no, that'd make the distinction almost pointless I think
eli_oat joined the channel
# 20:22 aaronpk IMO as soon as you have text both above and below a photo it's not a photo post anymore
# 20:23 sknebel Kaja << "add !mf command, people forget the 2 all the time"
# 20:23 Zegnat You have e-content, so that one should be fine, gRegorLove
# 20:24 Zegnat I think it only happens if the image is a direct child of the h-entry
# 20:24 sknebel so I guess many posts are "saved" by layouting divs and stuff ;)
# 20:24 Zegnat Well. Doesn’t have to be direct child, there can be a single :only-child wrapper in between
# 20:25 Zegnat <div class="h-entry"><div class="e-content"><p></p><img><p></p></div></div>
[jeremycherfas] joined the channel
# 20:27 Loqi sknebel has 19 karma in this channel (51 overall)
singpolyma and snarfed joined the channel
# 20:33 aaronpk hm i need to update my twitter POSSE to use the new reply thing wher eyou don't need to include ppls names in the tweet
snarfed, [eddie], [miklb] and tantek joined the channel
# 21:45 Loqi [sknebel] #34 Incorrect implicit name properties
# 21:47 sknebel Zegnat aaronpk gRegorLove: I feel like the implied "photo" parsing should be an issue against the mf2 parsing spec, yes/no?
# 21:49 gRegorLove Which part, specifically? Did we have a real world example of it giving uexpected /PTD?
# 21:53 sknebel gRegorLove: your article is only not an example because you used a figure tag around the img, right?
davidmead joined the channel
# 21:56 Zegnat gRegorLove probably has some more elements outside the e-content, so his will not trigger implied photo
# 21:56 gRegorLove No, because it's inside div class=e-content, and that div has sibling elements
# 21:56 Zegnat Yeah, assoon as there are siblings there it is off
# 21:57 gRegorLove Not opposed to a github issue on it, though, or discussion in #microformats at least. tantek can shed more light on the history of it, probably.
# 21:58 Zegnat Really it is a bit of an edge-case. I can come up with loads of theoretical problems, but hard to find actual examples.
# 21:58 Zegnat Also because I think very few people push what mf2 can do, most copy examples from wikis and those are already proven
# 21:59 sknebel right, a sibling for the author etc is quite likely. to the point where I wonder if actual photo posts ever have a matching DOM?
# 21:59 Zegnat <span class="h-card"><img src="" class="u-feature" alt="not my face, not my photo, just a nice picture of my house"><span class="p-name">Martijn</span></span> - would trigger a faulty implied photo property and clash with the feature property. But again, pretty contrived example.
# 21:59 sknebel I think the u-featured issue is the most likely to actually happen?
# 22:00 Zegnat More likely, yes, especially if people put that outside of e-content. (Do they?)
# 22:03 sknebel all sites I'm looking at have a footer of some kind on every post that'll kill the nested rule
# 22:04 gRegorLove notes seem most likely for that to happen with, since they'll either not have a p-name at all, or it's on the e-content element
# 22:04 gRegorLove If you post an article and have the p-name inside the e-content, that parses fine, but will look weird in readers (duplicate name showing up)
# 22:09 Zegnat I always have p-name inside my content (unless I don’t define one and set it to be equal the content)
KevinMarks_ joined the channel
# 22:31 sknebel interesting, nearly every page I look at uses <figure> tags
# 22:33 sknebel (now I'd love it if we had parser logs telling us which rules are actually used how often ;))
# 22:42 gRegorLove Zegnat: Interesting, why don't you want the title to show up on the second link there?
# 22:43 Zegnat gRegorLove: I have no CSS for headings, and most of the time don’t see a need for them on my site. Put incase someone is converting my site in their reader, they might expect a title in there. Or when converting h-feed to RSS.
davidmead joined the channel
# 22:44 Zegnat This theoretically also means they have a title to use for bookmark-of/in-reply-to/… posts
# 22:44 Zegnat Though most people don’t seem to go that far in the parsing for their target posts
# 22:45 Zegnat Would have loved to go without the DATA element there, but I wanted a more proper title then I could wrangle the text to be ;)
# 22:45 gRegorLove Yeah, that's something I'd like to do. I'm usually semi-manually making my like slugs "star-" + original slug. Some don't have the nicest slugs, though :)
# 22:45 Zegnat On the event I didn’t need data because I already write “Virtual Homebrew Website Club” in the text just the way I would make the title
# 22:49 Loqi [eli] It isn't exactly the same thing, but check out Battle Decks:
Think of it as improv karaoke powerpoint.
Contenders each prepare a 10 slide deck for their opponent to present sight unseen. The audience judges by show of applause based on continuity,...
# 22:50 Zegnat Now I am really off for bed :) Have a good HWC wherever you are gRegorLove :D
# 22:51 gRegorLove Could move from <data> to <title> if you move your h-entry to the <html> :) G'night!
# 22:52 tantek yet people usually put visible post name in the body anyway, hence better to put h-entry on body
snarfed, jjuran, leg and tantek joined the channel