#dev 2018-05-25

2018-05-25 UTC
tantek joined the channel
#
tantek
!tell snarfed I'm starting with trying a simple text request to enable the OAuth app "Bridgy" on a particular github org, we'll see if that's enough information for an org admin to figure it out.
#
Loqi
Ok, I'll tell them that when I see them next
#
tantek
manual POSSEing for now
#
tantek.com
edited /GitHub (+277) "/* Bridgy */ provide a clue as to how to resolve POSSE errors to particular repos"
(view diff)
#
snarfed
tantek: thanks!
#
Loqi
snarfed: tantek left you a message 6 minutes ago: I'm starting with trying a simple text request to enable the OAuth app "Bridgy" on a particular github org, we'll see if that's enough information for an org admin to figure it out.
#
tantek
snarfed, I'll keep A / B / C testing various "prose" requests to org owner(s) and see which get the issue resolved the quickest
#
tantek
til then manual POSSEing it is for restrictive orgs/repos
snarfed, eli_oat, tantek, [jgmac1106], ben_thatmustbeme, renem and [unoabraham] joined the channel
#
@madebydusk
↩️ Mate, that would be amazing! It's one of those things, that when you wrap your head around webmentions it just makes sense. Looking forward to it, for now I'll just use http://comment.sh on my articles and keep publishing to http://micro.blog
(twitter.com/_/status/999870387734437888)
snarfed, [eddie], iasai, KartikPrabhu and swentel joined the channel
#
@growdigital
↩️ Plenty eloquent 🙂 – the alternatives are too complex, folks are locked out. My brain melts just thinking about setting up #WebMention…
(twitter.com/_/status/999939551014551552)
tglobe joined the channel
#
@EatPodcast
↩️ I may be missing something subtle here, but if you have a photo as part of your h-card, that works pretty well in an #indieweb context. Depending on the receiver, it can show up in likes, reposts, webmentions etc.
(twitter.com/_/status/999943835454181376)
jeremycherfas joined the channel
#
@vincentlistens
↩️ Sorry I probably wasn't articulate! Using the h-card would be my preferred method. So when I sign in to "any_service" instead of it trying to grab a Gravatar etc I can use IndieAuth to verify my own details and push them across.
(twitter.com/_/status/999945210070208514)
iasai, [unoabraham] and [Vanessa] joined the channel
#
Zegnat
jeremycherfas, learn about a new technology every day: https://twitter.com/martijnvdven/status/999949082582974464 ;)
#
@martijnvdven
@vincentlistens There is (was? 😔) @libravatar, which defined a way for websites to discover selfhosted avatars based on email addresses exactly like Gravatar. Some Gravatar plugins may even already support it. Just need to be in the know about it. https://www.libravatar.org/
(twitter.com/_/status/999949082582974464)
#
jeremycherfas
But not for long "Libravatar will be shutting down on 2018-09-01"
#
Zegnat
But you can selfhost it. That’s the whole idea. Otherwise Libravatar would just be a Gravatar copy
#
Zegnat
The question is, how many plugins will still support it if the hosted version dies. Throwback to our HWC discussion about hosted Known (and other CMS)
#
jeremycherfas
Right. And how much would [people be willing to pay for something that isn't Gravatar and that probably would not be as widely available.
#
skippy
ahoy hoy
eli_oat, [jgmac1106], Kyle-K, [Vanessa] and [kevinmarks] joined the channel
#
[kevinmarks]
bridgy seems to have been confused by my twitter suspension - when I try to reauth it goes back to the auth page twice, and doesn't show new updates. [snarfed] - ping me if you want to debug together later
[miklb] joined the channel; Kyle-K left the channel
#
jeremycherfas
!tell aaronpk For some reason, after an update to Known, pictures from OYG no longer show up. The post is there, with tags, but no image. I reconfirmed by signing out, revoking the token, and signing back in again. Resending still has no image. Not sure where to start looking to debug.
#
Loqi
Ok, I'll tell them that when I see them next
[jgmac1106] and iasai joined the channel
#
swentel
hmm, is there an alternative for microformats to set a title, which doesn't get picked up by microformat parsers as 'name', but could be used as a title for say atom feeds ?
#
swentel
e.g. I'm lazy and would like to use granary to generate atom feeds
#
swentel
but, the titles for my 'notes' are quite clumsy
#
swentel
because they do not include p-name in my microformats feed
#
swentel
(as it should, unless I put p-name on the same wrapper as e-content as aaron is doing)
#
schmarty
swentel: i use granary for this and it looks like i use an explicit p-name when my post has no title
#
schmarty
i probably copied this from aaronpk :}
#
swentel
yeah, but that is on your e-content wrapper
#
swentel
so the title is kind of well, almost the same as the body
#
swentel
I could write it myself, but I'd rather use granary too :)
#
swentel
because in Drupal, every node does have an explicit title
#
swentel
s/have/has
#
swentel
going to read the micrformats spec
#
swentel
I hope there's some 'feed-title' class or so lol
snarfed joined the channel
#
swentel
snarfed, have you seen my issue by any chance I created on granary (and closed again?)
#
[jgmac1106]
This is what I was telling @gwg we should do with the note kind in WordPress "(as it should, unless I put p-name on the same wrapper as e-content as aaron is doing)"
leg and [eddie] joined the channel
#
sknebel
swentel: what kind of title would you want otherwise?
#
snarfed
swentel: hi! yes!
#
swentel
sknebel, well - https://granary.io/url?url=https://realize.be/timeline&input=html&output=atom - my notes have an internal title with #{1} - but they look a bit weird in atom feed
#
swentel
snarfed, yeah, I was trying to (re)figure out again whether I could still use granary, and have a nicer title as it generates now, any pointers greatly appreciated :)
#
swentel
s/was/am
#
sknebel
sure, that's not great. but if not something like that and not a manually choosen title and not the content, what would you use then?
#
swentel
sknebel, well, for my notes #32 would just be fine there, I can live with that (or maybe I change the titles to 'Note x'
#
schmarty
swentel: name is one of the "implied" properties that the parser will "figure out" if you're not explicit: http://microformats.org/wiki/microformats2-implied-properties
#
snarfed
...and implicit p-name isn't great in a bunch of ways
#
snarfed
it's improving, but mf2py (granary's parser) doesn't have all of the recent improvements yet
#
swentel
yeah, but for a 'note' post, p-name shouldn't exist (or be one the same wrapper as e-content)
#
swentel
hmm, I could do that of course
#
schmarty
swentel: you could use a hidden element to give an explicit p-name. for example, <data class="p-name" value="Note x">
#
sknebel
schmarty: which turns it into an article
#
snarfed
and re the feed title itself, i haven't investigated yet, but try including an explicit p-name in the h-feed (and outside entries)
#
sknebel
so indiereaders will think they have to show that as an useful title
#
swentel
if xray ignores that, that would be great, but I doubt that
#
swentel
snarfed, ok, will try :)
#
swentel
other than that it's working nicely, almost have the fediverse integration working with drupal
#
Loqi
[singpolyma] #145 Some weird results
#
sknebel
I follow a bunch of granary feeds and I haven't found the title being the name for notes being a problem
#
snarfed
common denominator is mostly bugs and missing features in mf2py
#
schmarty
sknebel: good point, post-type-discovery would pick that up as an article, then. i think combining e-content p-name is the most widely used method to get around this problem.
#
snarfed
recent mf2py release fixes some of them, i'm working on upgrading to that
#
schmarty
and consumers are expected to "dedupe" if name is a subset of content
#
snarfed
schmarty: yeah, which can fail in granary due to weak mf2py whitespace handling, a la https://github.com/snarfed/granary/issues/142
#
Loqi
[aaronpk] #142 Whitespace tags stripped from plaintext values
#
schmarty
swentel: that is very exciting! so folks in the fediverse can follow you?
#
swentel
schmarty, almost .. putting all the pieces together now
#
schmarty
snarfed: right! thanks for the links. i had vague memories of the implied p-name changes and mf2py updates for it.
#
schmarty
KartikPrabhu++ for mf2py updates
#
Loqi
kartikprabhu has 24 karma in this channel (202 overall)
[kevinmarks] joined the channel
#
sknebel
snarfed: do you have an example where that fails? even with bad whitespace handling, p-name and e-content should result in the same plaintext version?
#
sknebel
(hm, there was something with image handling I remember from aaronpk?)
#
Loqi
[aaronpk] #142 Whitespace tags stripped from plaintext values
#
sknebel
snarfed: but name and content are identical there?
#
snarfed
depends on whether you compare html or value i guess
#
sknebel
comparing html directly clearly doesn't work
#
sknebel
and with your own text-conversion of it, yeah, then you're tied to doing it *exactly* like the parser does
#
snarfed
it sounded like that comparison was at least part of the root cause there, but i don't remember all of the details
#
swentel
ok, I'm going to start overriding the node templates in Drupal
#
swentel
so the parsers have less trouble with these things :)
#
realize.be
edited /Webmention (+61) "/* Drupal */"
(view diff)
barpthewire, tglobe, [chrisaldrich], maingo, [keithjgrant], snarfed, KartikPrabhu, [cleverdevil], AngeloGladding and tantek joined the channel
#
Loqi
[cleverdevil] microformats2: Validate Microformats2 JSON with JSON Schema, and perform Post Type Discovery
#
@jgmac1106
Never really had a post go viral on LinkedIn (if 700+ could be considered viral) but my post on webmentions and credentialing is the most active post I have ever had: https://www.linkedin.com/feed/update/urn:li:activity:6405191090282074112 #OpenBadges #Edtechchat (https://jgregorymcverry.com/2893-2/)
(twitter.com/_/status/1000061824669233152)
#
Zegnat
Reading back a little. Yeah. I think the deduping rule is if p-name === e-content[value] (not e-content[html]), assume the author meant for there to be no title. But I couldn’t actually point you to the direct discussion on the wiki.
snarfed joined the channel
#
Zegnat
That should probably be cleaned up a little though. But is that the kind of title-note-overlap you were thinking about, swentel?
#
Zegnat
If Indie Readers follow that, your titles could be filled with just the first few words of the note text, followed with “...”. And readers should ignore it, while you can still have it in your generated feeds.
#
swentel
well, I used to have the title of my note in a p-name
#
swentel
but now it's on the same wrapper as e-content
#
swentel
but I'd be nice if could have a say 'feed-title' class or so (just a loud thought)
[chrisaldrich] joined the channel
#
swentel
come on mastadon, do your work .. :)
#
[jgmac1106]
Putting p-name in same wrapper as e-content is same thing Known does@
#
Zegnat
Yes. It is actually a bit of a workaround because until recently you couldn’t not have a name property. With the current mf2 spec, you should be able to keep the p-name away entirely for note posts.
#
Zegnat
But because the name property was required, the note type algorithm was invented so readers could decide when they did not want to display the title at all. This should theoretically mean you can define a (hidden) p-name specifically for things like RSS generation, that will not show up in Indie Readers.
#
swentel
Zegnat, oh but xray picked it up
#
swentel
even the hidden one
#
swentel
so I saw a title in monocle etc
#
Zegnat
(E.g. use <data class="p-name" value=""> with a value of “first five words of note …” and people should not be displaying it.)
#
Zegnat
It is possible that xray/monocle isn’t implementing the note type algo. Could file a bug for that
#
[jgmac1106]
Yeah with a title in Known name is an H2 class and their is no p name in post body just e content wrapper
#
@thorkon
Ist die #brid.gy-Brücke nach #Facebook kaputt? Help anyone?#webmentions
(twitter.com/_/status/1000071384570253312)
#
swentel
ooh, I always assumed xray would be very much on top of things
#
Zegnat
I don’t actually know who/what/where/when the note algo got invented/implemented. I only had a vague recollection of it existing.
#
swentel
but haven't tried the data tag yet
#
swentel
mayebe I should
#
swentel
and see what happens
#
Zegnat
Tag should not have any effect on the parser outcome. If it isn’t deduping titles, then it just doesn’t support that :(
#
swentel
ok, let me try
#
Zegnat
But sounds like something that might be worth implementing in XRay! Definitely encourage opening an issue for it to get that discussion started.
#
Zegnat
Hmm. Deduping must be happening somewhere, tanteks notes don’t get titles: http://xray.p3k.io/parse?url=http%3A%2F%2Ftantek.com%2F2018%2F140%2Ft1%2Fbaytobreakers-bonus
#
Zegnat
And he uses "p-name e-content" markup
#
Zegnat
But maybe only exact matches and not prefixes are deduped
#
Zegnat
swentel, there should be deduping, and elipsis titles should be supported by XRay: https://github.com/aaronpk/XRay/blob/master/lib/XRay/Formats/Mf2.php#L373-L385
#
Zegnat
(I should probably just have gone and checked immediately. Rather than taking you all along on my thought process.)
#
swentel
no wories
#
Zegnat
XRay seems to use similar logic when it is parsing RSS/Atom feeds. So I would make the first-five-words-followed-by-ellipsis my definitive recommendation :) If you see readers that do not implement this, file bugs with them.
#
snarfed
swentel: if you put a p-name in your h-feed, outside any h-entry, i think granary should convert that to the atom feed title
#
swentel
snarfed, yeah, been going through the code .. and it looks like it
#
swentel
the case it works for aaron is probably because he has no h-feed class on his homepage, and p-name is in his h-card
#
swentel
(at least I think so)
#
swentel
(been a while since I read python)
#
swentel
snarfed, yep, that's it
#
Loqi
woot
#
Zegnat
I am somewhat happy most parsing tools break on my website, so my markup only needs to be good /in theory/ rather than /in practise/ :P
#
Zegnat
I expect licit.li to fail
#
snarfed
connecting, not parsing
#
Zegnat
1 little switch on the server broke all of the tooling, hahaha
#
Loqi
ahahahaha
#
Zegnat
Well... except my own tooling. I can still parse my site
#
snarfed
but can you *connect* to your site
#
snarfed
"must be your hardware"
#
swentel
ok, atom feed title check
#
swentel
now, back to fediverse testing :)
#
Zegnat
It is because I use actual up-to-date libraries, snarfed :P
#
snarfed
ooooooh them's fighting words
#
Zegnat
Hahahaha
#
Zegnat
My local version of PHP makes use of my local version of curl. And that is the important part.
#
Zegnat
But granary runs on appengine and I assume you don’t control what is used to fetch external URLs?
#
snarfed
right. some parts yes, other parts no
#
snarfed
fwiw, it's *extremely* rare that i can load a web site in a browser or curl but granary can't
#
Zegnat
points at his blog
#
Zegnat
But yeah, a lot of web tooling in general is completely failing at retrieving my blog.
#
Zegnat
This gives me a little bit of joy. Especially as basically all browsers connect without issue so I am not locking people out at all
#
snarfed
the agony and ecstasy of indieweb. you get to choose for yourself how your web site works!
#
Zegnat
And of course I have a status as edgiest edgecase to defend!
#
Zegnat
Though I am bending the standards a little. nginx can’t (as far as I have been able to find) really do what I am doing in the proper way :(
#
snarfed
tantek may fight you for that
#
sknebel
Zegnat: indeed, your server is lying about what it supports
#
Zegnat
I was looking at caddy recently and it may be able to do what I do now without lying!
#
sknebel
so have some compassion for poor tools that just trust it to tell the truth :P
#
sknebel
fairly sure you can use HAProxy to achieve it too
#
sknebel
but I think we had that conversation alread
#
Zegnat
Yeah. But setting up an extra layer in front of the server, with all the other possible changes in the TLS settings, still doesn’t feel worth it to me.
#
Zegnat
Swapping to Caddy might be easier (and interesting in itself)
#
Zegnat
Another reason I have considered swapping to Caddy: I can run what is almost the exact same executable on my dev machine as I run on the VPS. Making the local environment match the server environment more closely
#
sknebel
you can run what is almost the exact same nginx executable too :P
#
www.boffosocko.com
edited /on_this_day (+348) "Twitter feature using advanced search"
(view diff)
[stefp] joined the channel
#
www.boffosocko.com
edited /Facebook_Master_Algorithm (+235) "/* Twitter Algorithmic Feed */ method around algorithmic feed using advanced search"
(view diff)
barpthewire, [kevinmarks], [jgmac1106] and [chrisaldrich] joined the channel
#
www.boffosocko.com
edited /Facebook_Master_Algorithm (+364) "/* Similar Algorithms */ Lindy Developer Twitter app for non-algoirthmic Twitter timeline; Keith Grant gist for Adblock"
(view diff)
[miklb] joined the channel
#
[miklb]
hey static site folks. How do you store your webmentions—JSON data files or markdown with YAML?
[jeremycherfas], gRegorLove and snarfed joined the channel
#
swentel
oooooooh
#
swentel
snarfed
#
swentel
like works :)
#
Loqi
[Ryan Barrett] Working on connecting Mastodon with the #IndieWeb! https://fed.brid.gy/ https://snarfed.org/indieweb-ostatus-bridge
#
snarfed
that's something
#
Loqi
does a happy dance!
#
swentel
can't see that it's me liking though
#
swentel
but it was def 0 before :)
#
swentel
interestingly enough, still got '202 response! If this is Mastodon 1.x, their signature verification probably failed.' back in the response
#
snarfed
oh mastodon, never change
#
swentel
so it's the same response for a reply, but something is going on there then I guess
#
swentel
let me try a repost
#
swentel
works too
#
swentel
but again, same response :)
AngeloGladding, KartikPrabhu and [Vanessa] joined the channel
#
Zegnat
https://www.passprotect.io/ - interesting browser extension by Okta, going to check out how it is intercepting passwords. Too bad Firefox in “coming soon”. Would’ve expected Firefox and Chrome extension code to be close to the same for this.
snarfed and tantek joined the channel
#
@thorkon
@zdfaspekte #aspekte #DSGVO: Es ist Zeit für #webmentions . Dann ist „der Weg zurück ins Web 1.0“ unnötig.
(twitter.com/_/status/1000124184650027011)
[jgmac1106] joined the channel
#
tantek
Anyone know if Bridgy Backfeed from FB has been impacted recently? I'm seeing RSVPs on a FB POSSE event, but none of them making it back to the original via Bridgy? http://tantek.com/2018/150/e1/homebrew-website-club-sf
snarfed joined the channel
#
tantek
good timing snarfed
#
snarfed
reads logs
#
Loqi
[armingrewe] #826 Seems Facebook mentions have stopped again
#
snarfed
only some though
#
tantek
snarfed, on my dashboard page however it is saying "Last syndication link found a month ago. " which has me concerned something worse broke
#
tantek
since I just posted the FB POSSE event yesterday
#
tantek
tried a "Poll now", and also tried Disable / Enable toggling
#
snarfed
yeah very possible. i haven't investigated yet
#
tantek
I mean there may be multiple problems, hence I'm trying to track down possibilities
#
snarfed
do you have an example post on your site with a synd link to a FB post of yours? ie not the indiewebcamp page's?
#
tantek
has a u-syndication link to a FB POSSE event
#
snarfed
bridgy pays attention to which FB page a synd link goes to
#
snarfed
the indiewebcamp page is https://brid.gy/facebook/688447421193285 , which doesn't have tantek.com in its profile (or didn't when it was last authed), so it doesn't look for synd links on your site
#
Loqi
Indie Web
#
snarfed
maybe add your site to the page's profile and re-auth it into bridgy
#
tantek
no this is always how it worked
#
tantek
You're right that if the creator of the event was the "IndieWebCamp" user/page on FB, then it wouldn't recognize it
#
snarfed
ah i see, got it
#
tantek
but when the creator of the FB event is me (on FB) then Bridgy would recognize the connection and send backfeed
#
snarfed
yeah bridgy isn't getting that event from the FB API
#
tantek
I thought it got it from my site?
#
tantek
from the u-syndication link from my event?
#
snarfed
it doesn't use synd links to look up API objects, no
#
snarfed
in general it's difficult/unreliable to look up FB API objects by bare ids, sadly
#
tantek
scroll down in that log and see how it found https://www.facebook.com/events/2026940924191951/
#
tantek
2018-05-25 00:30:26.790770 D processing permalink: http://tantek.com/2018/122/e1/homebrew-website-club-sf
#
Loqi
Homebrew Website Club SF — Special 502 Bad Gateway Edition!
#
tantek
2018-05-25 00:30:26.791020 D previously found relationship(s) for original http://tantek.com/2018/122/e1/homebrew-website-club-sf: [u'https://www.facebook.com/events/2026940924191951']
#
Loqi
Homebrew Website Club SF — Special 502 Bad Gateway Edition!
#
tantek
2018-05-25 00:30:26.858510 D u-syndication links on the h-feed h-entry: [u'https://www.facebook.com/events/2026940924191951/']
#
snarfed
yes, again, it found that *synd link*. it didn't find the *FB API object*
#
snarfed
ie, when it asked FB for events via https://graph.facebook.com/v2.10/me/events?type=created&fields=rsvp_status , i expect FB returned nothing
#
snarfed
probably means i need to prioritize https://github.com/snarfed/bridgy/issues/814
[miklb] joined the channel
#
Loqi
[snarfed] #814 facebook: redo app review for existing permission
#
tantek
hmm yes I did not get any FB backfeed on the previous (502) event either
#
tantek
this is the last event that Bridgy sent me backfeed for: http://tantek.com/2018/094/e1/homebrew-website-club-sf
#
Loqi
Homebrew Website Club SF — Special 404 Edition!
#
Loqi
Homebrew Website Club SF — Special 502 Bad Gateway Edition!
#
tantek
I realize 28 days is not a very narrow window :/
#
tantek
Ugh this? "Deprecated: / Endpoints that return data including /feed, /posts, /comments, and the RSVP list."
#
tantek
(under Events API)
#
snarfed
heh yes and my memory is awful. from https://github.com/snarfed/bridgy/issues/814 :
#
Loqi
[snarfed] #814 facebook: redo app review for existing permission
#
snarfed
"they've largely disabled the events API, including /me/events, which we use to get events and then their RSVPs. originally it kept working even after they announced that they'd disabled it, but it stopped working today. that API call now returns no data."
#
snarfed
i conflated that with non-event backfeed which evidently partially broke today too
#
tantek
goes to check the newsletter
[colinwalker] joined the channel
#
tantek
!tell schmarty any idea why your responses to this post are showing up with a scary dark figure in hat, sunglasses, popped overcoat icon? http://cathieleblanc.plymouthcreate.net/blog/indieweb/why-we-need-the-indieweb/#comments
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
[martymcgui.re] This Article was mentioned on martymcgui.re
#
schmarty
Haha that is weird. My guess is that my photo was not picked up.
#
Loqi
schmarty: tantek left you a message 1 minute ago: any idea why your responses to this post are showing up with a scary dark figure in hat, sunglasses, popped overcoat icon? http://cathieleblanc.plymouthcreate.net/blog/indieweb/why-we-need-the-indieweb/#comments
#
gRegorLove
Looks like Semantic Linkbacks plugin, user-secret.svg
#
gRegorLove
Or schmarty is a spy.
#
schmarty
Could be both
#
tantek
why was your photo not picked up?
#
tantek
hmm there's no "backfeed" section on /Facebook
snarfed joined the channel
#
tantek.com
edited /Facebook (+2569) "move how to backfeed from the [[backfeed]] page to here, move weaning further up, right after export"
(view diff)
#
tantek.com
edited /backfeed () "(-2234) move FB backfeed details to actual Facebook page"
(view diff)
[Natris1979] joined the channel
#
tantek.com
edited /Facebook (+462) "/* How to backfeed from */ Note recent limitations / challenges FB dev blog post about event API deprecations"
(view diff)
#
tantek
!tell snarfed I moved the FB-specific backfeed info from /backfeed to /Facebook#How_to_backfeed_from and added citations of the dev blog post you noted - please review - thanks!
#
Loqi
Ok, I'll tell them that when I see them next
#
tantek
maybe its time to figure out microsyntax for allowing people to RSVP via Twitter @-replies
snarfed joined the channel
#
tantek.com
created /websignin (+25) "r"
(view diff)
snarfed joined the channel
#
tantek.com
edited /Why_web_sign-in (+379) "/* But a domain is a big investment */ That's the point - only a small fraction need to do the iteratively improving"
(view diff)
[asuh], [miklb], leg and [Vanessa] joined the channel