#dev 2020-11-03

2020-11-03 UTC
gxt joined the channel
#
gRegorLove
Is Bridgy Instagram having issues for anyone? May just be temporary but in the logs I see "JSON script tag not found!"
#
gRegorLove
(for backfeed)
#
aaronpk
instagram--
#
Loqi
instagram has -2 karma in this channel over the last year (-3 in all channels)
#
[tantek]
It's definitely been one of the motivations for me to stop posting to IG since June.
#
[tantek]
posting anything persistent anyway
[chrisaldrich], deltab, geoffo, nickodd and velox joined the channel
#
gRegorLove
I manually POSSE there, but the backfeed is nice
geoffo joined the channel
#
[chrisaldrich]
gRegorLove, I manually POSSEd my first Instagram post since January and think I may be having the same problem with Brid.gy and backfeed. It's definitely not encouraging my return again anytime soon...
#
[chrisaldrich]
I think Ryan popped up over the weekend(?) and mentioned he was doing some maintenance...
KartikPrabhu joined the channel
#
gRegorLove
I'm working on an IndieAuth server for ProcessWire again, looking at the spec updates
#
gRegorLove
clients are supposed to send response_type=code now right?
#
gRegorLove
Noticed indielogin.com is sending response_type=id
geoffo joined the channel
#
gRegorLove
I'm guessing the server should accept id as a fallback, though
[tw2113_Slack_] and cambridgeport90 joined the channel
#
aaronpk
For backcompat yea
#
aaronpk
and I'm overdue for updating indielogin.com now
swentel, KartikPrabhu, cweiske, [KevinMarks], nolith, nolith[m], nickodd and [Denver_Prophit] joined the channel
#
[Denver_Prophit]
Quick question about `rel="u-syndication` : Can I put that as `<link rel="u-syndication" content="...">`
#
[Denver_Prophit]
in the head
#
aaronpk
If you're talking about the microformats value "u-syndication" then no that has to go inside the h-entry element for parsers to find it. If your html tag is the h-entry then it could work but that's a bit odd
#
aaronpk
but that said, rel= is not for the u-syndication property
#
aaronpk
you'd use rel="syndication" and that can be in the head because it's not MF2
samwilson, genehack and smacko[m] joined the channel
#
aaronpk
You're mixing up mf2 and rel values with that `rel="u-syndication"` markup
#
@DenverProphitJr
Working on Twitter Live Chats event schema. Now integrating http://Indieweb.org WebMentions using #microformats2 u-syndication will grab all conversations back to the webpage for display.
(twitter.com/_/status/1323639883542515714)
geoffo, vilhalmer, [tw2113_Slack_], jalcine[m], JK_na, hirusi[m], Pete[m] and nolith[m] joined the channel
#
[Denver_Prophit]
see tweet about. Left side of screen: front matter yaml
mayakate[m], jamietanna[m] and nekr0z joined the channel
#
[Denver_Prophit]
above*
#
aaronpk
rel=u-syndication does not do what you want
#
aaronpk
you are probably looking for `<a class="u-syndication">` http://pin13.net/mf2/?id=20201103153501785
#
[Denver_Prophit]
yes I see I need class="" to be read, right?
#
[Denver_Prophit]
ok so I need to do a for loop and almost a dozen anchor links to define the syndication.
smacko[m], [chrisaldrich] and Rixon joined the channel
#
jacky
[Denver_Prophit]: do you have a HTML parsing library?
#
jacky
That's what I've leaned on for things like this, versus explicit looping mechanisms
#
[Denver_Prophit]
Jacky: For jekyll?
#
jacky
oh no, for an Elixir project; my fault
#
jacky
didn't realize that was a Ruby setup
#
jacky
but I'm sure there's one for Ruby; which also should be okay in a build setup
#
[Denver_Prophit]
Jacky: Probably so! I'll add to my read list.
nolith[m] joined the channel
#
nolith[m]
Hello. aaronpk I was looking at one of your plogpost and I'm curious in what is that you collect under the "Other mentions" section?
#
nolith[m]
Replies, bookmark, reposts, and likes all have their own place above it
#
aaronpk
"other mentions" is when it can't be classified into one of the other types
jamietanna[m] joined the channel
#
nolith[m]
but are webmetions or the old pingbacks?
#
aaronpk
webmention/pingback are the delivery mechanism, not the content
#
aaronpk
the contents of the post at that URL exist regardless of whether a webmention or pingback are sent
#
nolith[m]
good point. so you parse the link, extract the mf2 and figure out what it is.
#
aaronpk
correct
#
aaronpk
and if it isn't a reply or like etc then it goes under "other mentions"
#
nolith[m]
I'm really new to all of this and I'm implementing my own system based on a static site generator.
#
nolith[m]
And that section on your site seems to be part of the answer I'm looking for. But only on the receiving side
#
nolith[m]
In my understanding what I'm missing is how can I inform you that I wrote an article that mention you but it's not a reply/repost/like?
#
aaronpk
if your post has a link to someone else's post, then you can send a webmention at that point already. it's up to the received to decide what to do with it
#
aaronpk
if you add microformats, then it makes it easier for the receiver to show it as a comment or a like
#
nolith[m]
Ok. so it could be a good implementation to scan for links in my own articles/notes and send webmentions to everyone supporting it.
#
aaronpk
that's generally the idea. although some people are more reserved about which links they send webmentions to, such as only sending webmentions to specifically posts they're replying to or liking rather than "all links in the article"
#
nolith[m]
I think this was not clear to me reading about webmentions on indieweb.org
#
nolith[m]
Thank for explaining
#
nolith[m]
Many of the "other mentions" on that post seems to be just people tweeting a link to your page. Is that something that brid.gy supports out of the box?
samwilson, mayakate[m], aviraldg, jalcine[m], nekr0z, fredcy_, JK_na, khimaros[m], Salt[m], Rixon, hirusi[m], smacko[m] and Pete[m] joined the channel
#
aaronpk
yeah bridgy will look for links to your website and send webmentions for those tweets
#
nolith[m]
thank you.
jeremych_ and gRegorLove joined the channel
#
[Denver_Prophit]
yay! Got the `u-syndication` working. Now, because I have more than one, I'm getting duplicate likes. I'll apply jekyll liquid uniq 😃
#
Loqi
giggles
nickodd left the channel
#
@DenverProphitJr
#webmention from http://indieweb.org is sort of working. I need to filter uniques. LOL! Overall, each #facepile links back to their twitter comment link. Each repost includes the content formatted as HTML from JSON. https://denverprophit.us/events/technical-seo/fiction-book-essentials-chat.html
(twitter.com/_/status/1323678247821971456)
[KevinMarks] joined the channel
#
[Denver_Prophit]
Can webmention.io pick up mentions from goodreads?
#
aaronpk
the question is whether goodreads can send webmentions
#
aaronpk
or if bridgy supports goodreads to send webmentions for it
#
aaronpk
webmention.io (and any webmention server) will accept webmentions from anything generally
#
[Denver_Prophit]
oh duh I forgot. I meant something like brid.gy
#
[Denver_Prophit]
I founded https://github.com/snarfed/bridgy/issues/905 from [snarfed]
#
Loqi
[jamietanna] #905 [Silo Request] Goodreads.com
jolvera, [tw2113_Slack_] and swentel joined the channel
#
jacky
I'm still interested in using micropub to make one's own site into a 'database' of sorts
#
jacky
this expands into more 'micro' activities like listening to music
#
sknebel
jacky: i.e. using it to submit scrobbles etc, not as posts necessarily but just as data for other pieces to use?
#
aaronpk
i keep debating whether to post scrobbles like the rest of my posts or if that's going to just flood my storage
#
jacky
that's what I'm afraid of (re: storage flooding)
#
aaronpk
i mean it's not really a scalability issue, but like i average 20-50 posts a day right now, but if i also include scrobbles it's going to 3x or 4x that
schmudde joined the channel
#
aaronpk
and my regular posts are going to start having URLs in the 100s at the end (my url format is YYYY/MM/DD/N)
#
schmudde
[Denver_Prophit]: what do you mean by "inside an imageobject"?
#
[Denver_Prophit]
Schmudde: Sorry I was referring to JSON+LD but link rel works
#
jacky
I'm thinking of modifying a particular post to add when I did listen to a song
#
schmudde
Got it. Thanks. petermolnar's solution seems to do the trick nicely!
#
[Denver_Prophit]
it does! I'm blown away. Never thought to do that.
#
[Denver_Prophit]
I use canva.com a lot to create images. I might just affiliate link canva for the "get this image on:" link muaha
#
schmudde
Great idea.
#
[Denver_Prophit]
I think you MAKE the creation public and then add your referral code.
#
schmudde
Well I'm using public domain images from the Metropolitan Museum in NYC. And I mark them CC0 in the footer to do my small part to properly attribute and raise awareness of the license.
#
schmudde
But I started thinking that my CC0 wasn't machine readable and it should be.
#
[Denver_Prophit]
`<meta>` is machine readable. Just like Dublin Core meta tags.
#
[KevinMarks]
rel=license is too, but it applies to the whole page - did you need the Schema to get google to accept it?
#
KartikPrabhu
any HTML tag is machine readable; the real question is whether anyone cares to parse and consume it
swentel joined the channel
#
schmudde
KartikPrabhu: right. But it looks like petermolnar is getting results in Google image search by putting the license in the caption. I would have expected it in the img tag.
#
schmudde
Creative Commons doesn't offer a best practice, unfortunately.
#
schmudde
I'm not an expert here, [KevinMarks], but it looks like he has some schema.org markup. https://search.google.com/test/rich-results?id=M4vPCh2pcgbEAd97VixvCw
#
petermolnar
I do have schema.org along with the rel=license
#
petermolnar
the more the merrier, right?
#
petermolnar
it's a <figure> which contains an <img> and <figcaption>; the meta is in the <figcaption>, so in theory, it'll be applied for the <figure> itself. In theory.
#
petermolnar
to avoid duplicating the data, I opted (again... sigh) for RDFa.
#
petermolnar
BUT, because there is a BUT: in theory, google will read certain embedded in-image meta
#
petermolnar
including the licence
#
petermolnar
that is also there on the images
[snarfed] joined the channel
#
[snarfed]
thanks for the heads up on broken bridgy instagram backfeed, all! scraping, whee. i’ll take a look, should be straightforward to fix.
#
[snarfed]
also disappointing, i have a daily test for IG’s scraping in my CI, which usually catches these, and correctly failed this time too, but i didn’t get notified. 😐 https://app.circleci.com/pipelines/github/snarfed/granary/373/workflows/de2f24af-520c-401e-992e-15faebc83d28/jobs/2420
[schmarty], djmoch and schmudde joined the channel
#
[snarfed]
fixed! feel free to retry.
#
[snarfed]
on an unrelated note, bridgy’s wm endpoint caching strikes again 🙁 https://github.com/snarfed/bridgy/issues/985
#
[snarfed]
torn. ideally bridgy would only cache endpoints per URL, not per domain, but that would bump its endpoint discovery rate up to ~100k URLs per day, which is...a lot
#
Loqi
[nolith] #985 No twitter webmentions for abisso.org
maxwelljoslyn joined the channel
#
Loqi
[jacky] I'm still interested in using micropub to make one's own site into a 'database' of sorts
#
maxwelljoslyn
I am also very interested in using micropub to deliver data to my site
#
maxwelljoslyn
Now that I am finally fully done transitioning from static files to a site that's an app
#
maxwelljoslyn
lahacker and I have discussed 'overloading' an MP endpoint, such as by sending it a specific tag, which causes it to create data but not create a post - Location return header be damned
#
maxwelljoslyn
s/tag/category
#
maxwelljoslyn
And it would be nice to reuse all my MP code for receiving, validating, storing posts -- rather than write another little service to handle e.g. "update my 'currently reading'" which isn't a post, it's just a bit of info that gets rendered as part of my Books page
swentel joined the channel
#
jacky
hmm interesting
#
jacky
I think I wanted to go and use mf2 as the query language of sorts
#
jacky
like if I wanted to pull only my likes
#
jacky
but in the case of books
#
sknebel
extend micropub as an API for a document database, in a way? :D
#
jacky
I could scan against properties to get some things
#
jacky
like if I want to see my progress in a book maybe using `read-of` to fetch a book by its ISBN (`isbn://$ID`), its author's URL or Wikipedia page
#
maxwelljoslyn
sknebel: more or less! extending micropub to do things besides 'make a post' so that I don't need to make a whole API for my site before storing things that don't deserve their own post
#
jacky
I think the best example of this would be having a page on my site that references a book
#
jacky
and from there, you'd be able to see my read progress, latest reviews, etc
#
jacky
all because they're children to that page
[tantek] and [chrisaldrich] joined the channel
#
maxwelljoslyn
sknebel: the idea is to go further and use MP as basically a universal API to instruct my site to do things totally unrelated to indieweb. e.g. imagine sending an MP with categories "nopost" and "lightbulb". MP endpoint receives it, says "aha, this isn't actually a post," ignores the content entirely and instead tells my IoT lightbulb to turn on
#
maxwelljoslyn
it's got nothing to do with posting, I just want one entry point for "hey website, do stuff"
#
maxwelljoslyn
it's got nothing to do with posting, I just want one entry point for "hey website, do stuff"
#
aaronpk
at some point that starts to look less like micropub and more like just using the indieauth part to get an access token and then having it be some other api
#
sknebel
right. I think it makes sense to consider what you actually get from the structure MP provides
#
[snarfed]
fetching and parsing mf2 from your site’s feed(s) is the straw man alternative, which gets you a decent ways, but is limited to public/published posts
#
sknebel
e.g. I think editing book database entries makes sense - it's creating and editing structured objects
#
[snarfed]
(unless you add in AutoAuth etc)
#
sknebel
but overriding categories just to do stuff unrelated to actual posts feels odd
[KevinMarks] joined the channel
#
[KevinMarks]
not necessarily - Publ has private posts
#
aaronpk
if you really wanted to stretch it you could say that a room is a "post" and various lights are properties of the post and then use micropub edits to change their state
#
maxwelljoslyn
snarfed, I don't follow - do you mean 'make a post as normal, then have some other thing consuming those posts and taking action'?
#
maxwelljoslyn
aaronpk: LOL
#
maxwelljoslyn
my body is ready
#
[snarfed]
maxwelljoslyn [KevinMarks] heh sorry if i was unclear, i was a bit behind. for the “query my site for existing posts” part, i meant fetching HTML from your site’s h-feed(s) and parsing its mf2. (as the alternative to new micropub features for that)
#
maxwelljoslyn
"welcome to the part of my website where you can take a virtual tour of my house with real-time up to date lights. log in to turn them off and bother me"
#
[snarfed]
(more background: https://snarfed.org/posting-to-the-indieweb-from-your-phone#crawl …but it’s really just a straw man, not a serious proposal)
#
maxwelljoslyn
I understand now snarfed thanks. that makes sense
#
[KevinMarks]
feeds that show different things to different people are certainly doable, at least with an API based site
#
[snarfed]
right! hence my mention of AutoAuth
#
maxwelljoslyn
Part of this woolgathering stems form my reluctance to make a lot of posts but that's also becaues my old rickety SSG approach had no concept of "public/unlisted/private" and put everythig into 1 feed. now with more control I could just embrace the indieweb "everything is a post" ideal and just have a bunch of posts I'm not displaying (but which Hypothetical Light Controller is still watching for)
#
maxwelljoslyn
reading autoauth now, unfamiliar.
#
aaronpk
it's still...very rough
#
maxwelljoslyn
good, one less thing to lust after
#
sknebel
I haven't even found the calm time to incorporate the results from the last session :(
#
[tantek]
goodness sakes there was a lot of misinfo shared about license stuff and meta markup
#
[tantek]
what is Dublin Core
#
Loqi
Dublin Core (Dublin Core Schema) is a small set of vocabulary terms that can be used to describe digital resources (video, images, web pages, etc.), as well as physical resources such as books or CDs, and objects like artworks https://indieweb.org/Dublin_Core
#
[tantek]
yeah that's not upfront enough about debunking
[tw2113_Slack_] joined the channel
#
[tantek]
tl;dr, Dublin Core is nearly 100% waste of time. There are zero actual consumers of Dublin Core in HTML on the web. Anyone telling you to add it to your HTML is wasting your time.
#
[tantek]
It's the classic example of metacrap
#
[tantek]
that dfn of Dublin Core has zero specific relevance to indieweb
#
petermolnar
fully agreed; see https://twitter.com/RubenVerborgh/status/1092029740364587008 - my tweets are deleted in that thread, as any tweet of mine beyond 3 months age, but basically I asked him to show an example that consumes DC. I was not given one.
#
@RubenVerborgh
@petermolnar @csarven I'm sure there are many systems relying on DC (there is a yearly Dublin Core conference after all, where I see many librarians and archivists). FOAF, less so indeed. The majority of apps still needs to be built, so that's why creating a good developer experience is important.
(twitter.com/_/status/1092029740364587008)
#
petermolnar
the part I'm curious about if that anything RDF (DC, FOAF, AS, etc) can, in theory, be consumed by something that respects it, right?
#
petermolnar
So the problem is that nothing does, or that most, if not all tools are limited to certain vocabularies, eg Google to schema.org
#
petermolnar
btw, [tantek] you are incorrect on one statement in -chat: there are things consumed from <meta>: Open Graph and Twitter tags.
#
[tantek]
yes that's true that things consume OGP, TwitterCards from meta (and even meta name=description by Google search indexers), and in -chat I was specifically referring to no consumers meta DC from HTML
#
[tantek]
I don't think I ever claimed there is zero consumption from <meta> in general (as that's trivially provably false 🙂 )
#
petermolnar
aye, I misunderstood the context
#
[tantek]
no worries, it happens in text chat often, or I may have spoken ambiguosly myself
[kimberlyhirsh] joined the channel
#
[tantek]
it's been a while, but perhaps this blog post / dialog between rhiaro and me is a good example of an obscure corner of Dublin Core: https://tantek.com/2015/079/b1/dublin-core-application-profiles
#
Loqi
[Tantek Çelik] Dublin Core Application Profiles — A Brief Dialogue
[fluffy] joined the channel
#
[tw2113_Slack_]
i have never heard of Dublin Core