#dev 2021-01-13

2021-01-13 UTC
jeremycherfas, [KevinMarks], [fluffy] and strugee joined the channel
#
jacky
^ I feel like most people are talking about a library for webmention.io versus a library for interacting with a Webmention endpoint
#
jacky
which makes sense because the spec gives us 'half' of what one would need (a means of retrieval as well)
#
jacky
I've been thinking about if it'd be worth it to have endpoints also spit back out information about Webmentions tho
#
jacky
I feel like this might be a minority thought tho
#
aaronpk
this sounds familiar
#
jacky
I might not be using the right words
#
[KevinMarks]
you mean like fluffy's client script?
#
aaronpk
if webmention endpoint services all used the same api for retrieving them it would be convenient
#
jacky
aaronpk: right
#
aaronpk
and now that we have a reader spec, i'd probably recommend using the same syntax
#
jacky
[KevinMarks]: no - like if the endpoint had a uniform way of retrieval, similar to how Micropub has `q=source` (in one way)
#
jacky
wait there's a spec for readers?
#
jacky
I have missed quite a bit
#
jacky
what are readers
#
Loqi
A reader in the context of the indieweb is the portion/feature integrated into an indieweb site that provides a way to read content from other indieweb sites and respond (like, comment, repost, etc) inline on the site itself directly while reading posts https://indieweb.org/readers
#
aaronpk
microsub?
#
jacky
oooh ofc
#
jacky
peeks again at /Microsub-spec
#
jacky
yeah tbh
#
jacky
actually
#
jacky
I'll write out some thoughts on the wiki
#
aaronpk
plus webmention.io's API already is jf2 like microsub
#
jacky
I forget the mime type for jf2
#
jacky
probably (application/json+jf2)
KartikPrabhu joined the channel
#
jacky
oh that's official,officla
#
aaronpk
well i don't think it's ever been submitted to iana
#
jacky
oh okay
#
jacky
eh screw them lol
jeremycherfas joined the channel
#
aaronpk
jacky++
#
Loqi
jacky has 29 karma in this channel over the last year (101 in all channels)
[Raphael_Luckom] and jeremycherfas joined the channel
#
jacky
I hope that's an okay-ish approach
#
jacky
I didn't want to add _too_ much to it
#
[Raphael_Luckom]
when you say "a case where a site returns a specific feed for a particular reader (or a set of them)" -- are you referring to like, me returning HTML when I see a user-agent that wants to consume HTML, or does "reader" mean "permission set" like "all the mentions raph is allowed to see"?
#
[tantek]
oof I'm still trying to figure out how/what to store of the different aspects of a received webmention, since there's so many different kinds of data there (data received via the webmention, data *retrieved* via getting the source URL (including redirects), resources in that data like images, etc.)
#
[tantek]
having flashbacks to 2019 IWS conversations with [schmarty] and then with jacky near the middle/end of demo day
#
[tantek]
point being some of this data can be purged (it's "just" a cache) and some can't, so it lends itself to different storage policies
#
aaronpk
a lot of that is documented on post type discovery and h-entry now though
#
[tantek]
uh not really
#
aaronpk
and i already went through the exercise of storing the raw HTML to be able to re-parse them if i want, and it's just not practical, it ate up 1gb of storage real quick
#
[Raphael_Luckom]
interesting, I had considered going that way, now I probably won't
#
[tantek]
like the URL of the source vs redirected URL (at the time of first verification) vs raw HTML vs mf2json vs images
#
aaronpk
my approach is "store enough to show the response"
#
[tantek]
which feels more like a cache
#
aaronpk
which is better than not storing anything *ahem*
#
aaronpk
considers turning off tantek's webmention.io account to light a certain fire
#
[tantek]
gets a glass to pour one out
#
jacky
I'm very open to ideas on this
#
[tantek]
I'm guessing it's still mostly Bridgy, Bridgy from Twitter, Bridgy from Twitter @-mentions (not replies)
#
[tantek]
jacky, the conclusion I came to was a layered approach, driven by what aaronpk said, enough to display
[snarfed] joined the channel
#
[snarfed]
actually no bridgy stopped seeing and sending @t spam mentions a while ago
#
[snarfed]
no clue why
#
jacky
[Raphael_Luckom]: I'll expand on that case but this is like if, in the flow of authorization to one's social reader of choice, that they can set a particular 'permission' so that when it uses that same token to talk to your Webmention endpoint, it'd give it a particular feed
#
[tantek]
uh, snarfed, that first tweet result doesn't mention me at all. what's going on
#
jacky
my thinking around this was making a difference for something like indiebook to only get notifications about books
#
jacky
but not anything else
#
jacky
it's a very specialized case, now that I think about it
#
@lavigi
@trishankkarthik The Indie Web concept of a few years ago. "IndieWeb" https://indieweb.org
(twitter.com/_/status/1349083737007546380)
#
[snarfed]
[tantek] it mentions indeweb.org, which is in your twitter profile
#
[tantek]
now actually 🙈
#
[snarfed]
but doesn’t matter, that only makes it try to send to indieweb.org
#
[tantek]
more like a double 🤦‍♂️
#
[snarfed]
still follows basic mention rules etc. the goal really is to make silos do a bit of normal indieweb, not forward wms blindly
#
[tantek]
gets distracted rewriting his twitter profile
#
jacky
tbh one could just return a list of URLs (not even having to do all of the work on their server)
#
jacky
because then the client can handle the work of parsing the mf2 needed (and can even pass in extra data if things like authorization is needed)
#
jacky
[tantek]: re: presentation, your comments are making me think on it more and I think it'd be the above (like either Microsub timeline response: https://indieweb.org/Microsub-spec#Retrieve_Entries_in_a_Channel or just a list of URLs)
#
jacky
oh actually using the Microsub timeline flow, we can get free pagination
#
jacky
my ultimate goal with this is to 1: encourage more implementations to provide what people get from webmentions so that 2: we can normalize client libraries and make it easier to provide a feedback-centric interface
#
jacky
like it would be _amazing_ if we could have a community-backed implementation of a near-drop-in replacement for something like Disqus and it'd work regardless of the backend you use because it's the same interface
#
aaronpk
yeah i like that idea
#
aaronpk
also i want fewer people relying on webmention.io
#
aaronpk
in 2020 i worked on getting the indiewebcamp finances moved out of my LLC, time to de-aaronpk more things around here
#
jacky
plurality++
#
Loqi
plurality has 2 karma in this channel over the last year (8 in all channels)
#
jacky
soon come tbh
#
jacky
I really think that this would make it easier to pull away from wm.io
#
jacky
_and_ convince more JAMstack-y folks to try it out with their own systems (like actually attempting to implement Webmention all of the way through)
#
aaronpk
the fact that so many people have been writing posts about how to get webmentions on their websites means that's definitely a good next place to start
#
aaronpk
those posts should start out with "find (or build) something to receive webmentions" instead of "go sign up at webmention.io"
#
[tantek]
jacky, you're connecting the retrieval of webmention information with the reader use-case is exactly the same connection I made when for the "aggregation" axis of IndieMark when I last thought that through many years ago, note brainstormed "Levels" 5-6,6 (I'm not sure if those were two options for 6 or a typo for 6,7) https://indieweb.org/IndieMark#aggregation
#
[tantek]
it's all part of "getting & storing other people's stuff"
jeremycherfas joined the channel
#
[tantek]
which I summarized as "aggregating" as an umbrella term
#
[tantek]
note this is written from a developer perspective, and the levels are supposed to be increasing in both relative difficulty, and you're supposed to be able to use at least some of what you built for earlier levels as building blocks to build the higher levels
#
[tantek]
though the *result* of supporting each consecutive level of an axis is a user-visible benefit on your site
#
[tantek]
retrieving, storing, and *displaying* other people's stuff
#
[tantek]
I still think that "area" is one of the biggest challenges to get "right" both in terms of protocols, and more importantly social policies
ShadowKyogre joined the channel
#
[KevinMarks]
I tried to make mention.tech match webmention.io by returning jf2 and got fluffy's script to work with it, but we could iterate on that to make it more formal
#
Loqi
Tumble log like it’s 2005
hoschi joined the channel
#
[Raphael_Luckom]
thanks jacky, that makes sense
hoschi and [tantek] joined the channel
#
jacky
[tantek]: aggregation is a good term!
[jeremycherfas], jeremycherfas, [chrisaldrich], [fluffy], hoschi, nickodd, KartikPrabhu, schmudde, swentel, sebbu, jeremych_, jeremy, [KevinMarks], ShadowKyogre, [schmarty], [Raphael_Luckom], jolvera and [tantek] joined the channel; ShadowKyogre left the channel
#
[tantek]
[KevinMarks] you may have opinions on this issue regarding how to make it explicit that hidden content (like in summary/details) should still be found via Fragmentions (and find in browser UI etc.) https://github.com/w3c/csswg-drafts/issues/5595
#
[tantek]
(FWIW I don't think CSS is the right way to do it, but it is a problem worth solving IMO)
[KevinMarks] joined the channel
#
[KevinMarks]
I will have a look. Its an issue with regular fragment id's inside details too
#
[tantek]
that's a good point
#
[KevinMarks]
If we do want to expand details to a tabbed ui model, we need to think about these kinds of state
#
KartikPrabhu
I have a fragmention-aware expandable element on my site
jamietanna joined the channel
jeremycherfas joined the channel
#
KartikPrabhu
I think the scrolling seems to be a bit broken so it over scrolls, but the expandable element does expand if you link to something inside it with fragmention
j605 joined the channel
#
[tantek]
wow KartikPrabhu! can you add that as an example to that issue ^ ? and feel free to be explicit that you are supporting Fragmentions (link to spec) which predate ScrollToTextFragment
#
[tantek]
because too many people have now *only* heard of ScrollToTextFragment
jeremycherfas joined the channel
#
KartikPrabhu
that issue is discussing something about "search" no?
#
jacky
grumbles against fragmentions
#
KartikPrabhu
fragmentions are nice jacky
jeremycherfas joined the channel
#
jacky
grumbles louder
#
jacky
I feel like it clutters the heck out of a URL
#
aaronpk
the url is just the mechanism for sending the webmention
jeremycherfas, j605 and swentel joined the channel
#
jacky
ah I should be more clear - the whole hash-y thing in the URL is what throws me off
#
jacky
ah I should be more clear - the whole hash-y thing in the URL is what throws me off
#
jacky
but I think I also don't use it a lot so it's not stuck on me
#
jacky
chrome-only stuff
jeremycherfas joined the channel
#
KartikPrabhu
it works on firefox too which is what I use
#
KartikPrabhu
it does need an external JS on the page though
leg joined the channel
#
@bmann
I'm slowly having more Protocols > Products discussions, which needs to turn into a blog post at some point. Some of the pieces are there, and "defacto" standards like a markdown file with simple YAML front matter. But at a meta layer, Webmentions or Web Annotations link across
(twitter.com/_/status/1349434667334725633)
lanodan joined the channel
#
@ochtendgrijs
↩️ (Eigen website, die dankzij Webmention sowieso gewoon met andere sites kan communiceren. Kan andere[n hun compatibele] blogposts liken, beantwoorden en reposten dat het een lieve lust is, en omgekeerd, zonder dat daar een centrale partij aan te pas komt.)
(twitter.com/_/status/1349436805683281920)
jeremycherfas, schmudde and ShadowKyogre joined the channel; nickodd and ShadowKyogre left the channel
jeremycherfas, [Raphael_Luckom] and [tantek] joined the channel; ShadowKyogre left the channel
#
lahacker
aaronpk snarfed https://github.com/angelogladding/WebExtension-IndieAuth works for me (custom server) and maxwelljoslyn (indieauth.com)
jeremycherfas joined the channel
#
lahacker
https://github.com/canopy/liana shows it in a working extension
[snarfed] joined the channel
#
[snarfed]
lahacker++ thanks!
#
Loqi
lahacker has 4 karma in this channel over the last year (7 in all channels)
#
[snarfed]
any chance you could extend it to expose the auth code? i need to verify on the server side
#
lahacker
yeah it grabs an access token in the end
#
Loqi
[angelogladding] WebExtension-IndieAuth: WebExtension support for IndieAuth sign in
#
[snarfed]
right, but i’d need the auth code, not the access token
#
lahacker
oh.. for what?
#
[snarfed]
like i said, i need to verify the auth code on the server side
#
[snarfed]
my backend needs to verify that its client is authed for a given domain
#
lahacker
is that part of the spec?
#
[snarfed]
totally ok if it’s beyond what you plan to do
#
lahacker
i'd just like to know if i should be doing something with it in my server code
#
[snarfed]
oh no not necessarily. use case dependent