#dev 2019-04-11

2019-04-11 UTC
jjuran, [jgmac1106], stevestreza, ichoquo0Aigh9ie, dougbeal|mb1, [kenbauer] and KartikPrabhu joined the channel
#
GWG
If you are always storing location, how do you know when to show it or not?
eli_oat, [kenbauer], [fluffy], cweiske and ingoogni joined the channel
#
Zegnat
Sounds like a UI question? E.g. aaronpk is always showing his location through his background map image
[audioundergroun, [kevinmarks], swentel, gRegorLove, GWG, KartikPrabhu and [Rose] joined the channel
#
[Rose]
GWG: How about store a location visibility property? In the case of your plugin, perhaps "default", and then "public" and "private" would be the set values, so if I change my default setting the posts that didn't get an explicit setting before go with the flow.
#
Loqi
Ok, I'll tell them that when I see them next
#
ingoogni
!tell jgmac1106 browser that reads and writes, like Amaya (dead)? https://www.w3.org/Amaya/
#
GWG
[Rose]: I do that now. The absence of the property means the default is used. But there appears to be an issue with Micropub
#
GWG
Which means checkins from OYS aren't showing a location
[Rose] joined the channel
#
[Rose]
Oh, hmm
#
GWG
The other option is to use the logic that if the location comes from Micropub without the visibility property to assume public.
#
GWG
Or to set checkins to always be public
krychu joined the channel
#
GWG
I think I may have to stop my other project before I go back to webmention stuff
#
GWG
And fix this location problem
#
@jgmac1106
↩️ @francesbell do you have an image I can make an #openbadges and a webmention badge for #femedtech #oer19 (https://quickthoughts.jgregorymcverry.com/s/1lRBCH)
(twitter.com/_/status/1116306087479795712)
ingoogni and eli_oat joined the channel
#
GWG
Also got a bugfix to Micropub I may push if the reporter doesn't find others in a day or so
#
GWG
I need more hours
[kevinmarks] and [eddie] joined the channel
#
[eddie]
GWG I would definitely think a checkin post type via Micropub that isn’t a private visibility should show the location by default
[grantcodes] joined the channel
#
Zegnat
GWG, let me know when you fix that bug that takes up all the hours of the day. I am experiencing that too
#
GWG
Zegnat, what repository do I file the issue in?
#
Zegnat
sees himself out
#
GWG
I hope no one forks that repo
#
GWG
[eddie]: I have never used post type discovery to set location visibility
#
GWG
Micropub doesn't do PTD, nor does Location.
#
GWG
Modular pieces get confusing
#
GWG
So it is for me not only a problem and a solution, but where it goes
#
[eddie]
Yeah... I guess that’s tricky
#
GWG
I have a file in each plugin for integrations with another plugin
#
GWG
If installed
#
GWG
Should I assume all Micropub posts with a location property that don't explicitly set location as private are public?
#
GWG
This is more a general behavioral assumption that I would want to codify into the location visibility proposal
#
Loqi
[dshanske] #16 New Property: location-visibility
#
GWG
Wait, I did say that already
#
[eddie]
Yeah that makes sense and it matches the existing “visibility” behavior
#
[eddie]
And that doesn’t mean someone can’t set an app’s default location visibility to private, so that every request sends private
[stefp] joined the channel
#
[eddie]
So I don’t think it’s a privacy concern. I think that’s the right step from the spec side and the rest is about UI
#
GWG
[eddie]: All I need is for Indigenous for iOS to support location visibility
#
GWG
Or any client other than Indigenous for Android
#
GWG
Or any other endpoint
#
[eddie]
Makes sense. It’s on the list. Albeit, I’ll be honest, not SUPER high. There are some larger more broad scope issues within Indigenous for iOS in terms of usability that are higher priority
#
GWG
That and query for supported Properties
#
GWG
[eddie]: Just hoping it will catch on.
#
[eddie]
I DEFINITELY need to get some of the queries in there like supported properties, post types, etc
#
GWG
You could be the first client to support querying for queries and properties
#
GWG
No one has yet
#
GWG
Though I may try another PR to Quill someday
#
GWG
No one is doing profile in Indieauth support except Quill and WordPress I think
#
GWG
I also want to have expiring tokens, but no client supports them
[Rose] joined the channel
#
[Rose]
GWG what do you need? I could fork Quill, add the features and open a PR.
#
[Rose]
And I would then also have my own install of Quill which I want.
#
swentel
Drupal does too ;)
#
swentel
(the profile part)
#
GWG
swentel, it does
#
GWG
Good
#
GWG
Missed that
#
GWG
[Rose]: Just want to add support for some experimental ideas
#
swentel
yeah, have an issue open in indigenous to look for it if it exists
#
GWG
So there are more uses for it
#
[Rose]
If you could throw me a list then I can put it in my task manager.
#
swentel
there's a lot I could do actually which are supported by drupal/wordpress I think: pkce, queries and properties, indieauth profile
#
swentel
also token revocation
#
GWG
[Rose] : Will try to do that
#
swentel
I'd LOVE to see expire/refresh tokens btw too
#
GWG
The issue with refresh tokens is what do we do for backcompat?
#
GWG
Most clients don't support them
#
GWG
Which means that those clients will start failing
#
[Rose]
How do they handle an invalid token? Could that be used?
#
GWG
Not sure if oauth2 has a solution
#
GWG
[Rose]: Poorly in my experience
#
GWG
OYG and OYS just stop and you have to notice
#
swentel
of course, if the token is invalid, I could just restart the auth screen
#
GWG
For a client, it would just fail, but even Indigenous for Android didn't explain why well
#
[Rose]
Maybe expiring tokens would only be given to clients which explicitly say they support it?
#
GWG
But how do you say that?
#
[Rose]
In their config that can be called?
#
GWG
Also, subject to abuse if you allow clients to always get long lived tokens
#
[Rose]
I wouldn't see this as a long term solution, just as an interim one.
#
[Rose]
The same config that tells you where the media endpoint is for clarity.
#
sknebel
[Rose]: that's server config, not client config ;)
#
[Rose]
Hmm, valid point.
#
swentel
well, it could be in the q=config response, 'allows_refresh' or something like that
#
[Rose]
That's what I meant. Doesn't Q config also say where the media endpoint is?
#
GWG
That's Micropub, not Indieauth
#
GWG
Wrong layer
#
[Rose]
Darn.
#
[Rose]
I'm having one of those days. But I made it to the airport!
swentie, swentel, [schmarty] and [eddie] joined the channel
#
[eddie]
Yeah by time you do q=config you already have a token 🙂
snarfed, [jgmac1106], [schmarty], sebsel, [dougbeal], jackjamieson and ingoogni joined the channel
#
@johanbove
Got my first "stars" and "retweets" from a post written on #Known and #POSSE syndicated to the #Twitter "silo". Very exciting to have official joined the #indieweb. The whole process feels like magic thanks to #bridgy (https://brid.gy/) (https://social.johanbove.info/2019/04/11/got-my-first-stars-and-retweets-from)
(twitter.com/_/status/1116372536214867968)
[stefp], snarfed and leg joined the channel
#
snarfed
i have a narrow niche use case for the proxy html+mf2 pages that bridge services like bridgy and bridgy fed render for silo posts
#
snarfed
i want webmention receivers to fetch and handle them normally, but it might be nice for browsers to get redirected to the silo pages themselves
#
snarfed
i could use an HTML meta redirect tag, which in practice i expect would do what i want
#
snarfed
the question is: does that sound like an ok idea?
#
snarfed
example: it'd be nice if clicking on the mastodon likes and reposts in https://beesbuzz.biz/blog/7129-Memories#webmentions went to the mastodon post, instead of bridgy fed's ugly proxy pages
#
snarfed
a counterargument is that there maybe some meta redirects in the wild that webmention receivers should actually handle and follow. eg https://github.com/snarfed/bridgy/issues/253
#
Loqi
[snarfed] #253 support HTML meta http-equiv=refresh redirects
#
snarfed
(the webmention spec says discovery should "follow redirects," and links to https://fetch.spec.whatwg.org/, but i haven't found anything on HTML meta tag redirects in either doc)
#
sknebel
I seem to remember some library somewhere actually did handle meta tags as redirects, but I'd not commonly expect it
#
sknebel
although a compromise might be delayed redirect?
#
snarfed
hmm! how would that work? if wm discovery code followed meta redirects, i expect it wouldn't care about the timeout, right?
ingoogni joined the channel
#
sknebel
I feel like I remember code making a difference between an immediate redirect and a delayed one
#
sknebel
but certainly in need of study
#
snarfed
helpful first step might be to say explicitly whether meta redirects should be followed, in either the wm spec or fetch spec or wherever. cc aaronpk
[schmarty] joined the channel
#
[schmarty]
!tell [manton] a bunch of my posts at micro.blog/schmarty show no content, which is true that they have no content! they have u-bookmark-of, u-read-of, etc. makes sense that m.b doesn't support all of those, but i *do* include a p-summary that could be displayed.
#
Loqi
Ok, I'll tell them that when I see them next
#
[schmarty]
i'm trying to find where this has been discussed before and having trouble.
#
[schmarty]
ohohoho, good call!
[manton] joined the channel
#
[manton]
[schmarty] I was just looking at this. Micro.blog is reading your JSON Feed (converted from Granary). It does have a summary but the content is blank, which is really unusual.
#
[manton]
The spec says: "content_html and content_text are each optional strings — but one or both must be present."
#
[schmarty]
and thanks for that catch! it seems the issue i need to make sure granary picks up my summaries
#
[schmarty]
forgot that m.b is taking in JSON feed and not mf2 feed 😬
#
[manton]
Yeah, I think in this case the feed is technically valid but it seems weird not to have anything in the content. Micro.blog could use the summary instead... But why is there a summary if there's nothing being summarized? 🙂
#
[schmarty]
oh, interesting, okay. hmm.
#
[schmarty]
i guess the issue here is that summary exists on JSON feed
#
[schmarty]
and seems reasonable to map over
#
[schmarty]
but mf2 summary can (should?) be used as a fallback when there are unexpected mf2 properties, so i'd expect granary to map that to content_text.
#
snarfed
sounds like granary's emit invalid jsonfeed. should it maybe be stricter on jsonfeed output and fail (return http 400) if there's no content?
#
snarfed
or fallback, yes
#
[schmarty]
it's tricky. 🤔
[grantcodes] joined the channel
#
[grantcodes]
snarfed: You could use a js redirect?
#
[grantcodes]
Don't know if that makes sense but no scraper would follow it
#
snarfed
[grantcodes]: hmm true!
#
[manton]
[schmarty] What do you do in an RSS feed for these types of posts?
swentel joined the channel
#
[schmarty]
[manton]: i use granary for that as well. for RSS it looks like summary becomes the item description. atom feeds end up with empty content.
#
[schmarty]
time to own my feeds!
#
[schmarty]
unless snarfed would be interested in taking PRs to have granary fall back to summary if content would be empty in JSON feed and/or ATOM.
#
snarfed
PRs always welcome!
#
snarfed
although technically the atom content feed is optional, so i'm not sure i'd add the fallback there
#
snarfed
er, the atom content field
#
snarfed
fallback (or error) for jsonfeed sounds good though
#
[schmarty]
oh that's interesting re: content in ATOM feed. is there an expectation that granary does something like post-type-discovery to determine the content from an html+mf2 source?
#
snarfed
somewhat, yes, but that doesn't currently include summary
ingoogni, KartikPrabhu, [Rose], [dougbeal], [schmarty], [kevinmarks], eli_oat and [eddie] joined the channel
#
GWG
aaronpk, the expiring/refresh token question...is there a precedent in Oauth2 for how we move from non expiring to expiring?
#
aaronpk
from the app or server perspective?
#
aaronpk
in oauth, apps pretty much always have to be prepared to handle expired tokens and do the refresh flow
#
GWG
aaronpk, but since no Indieauth client currently supports it to my knowledge, what will be the concerns if the server does?
#
aaronpk
the clients will treat it as a generic error when trying to post, and what happens next depends greatly on the client
#
GWG
Which is a problem. Most don't handle it well
#
aaronpk
if an app doesn't handle it well, the "fix" that the user can do is to log out and back in again
#
aaronpk
which is not the worst thing in the world
#
aaronpk
like right now in quill it would show an error message if you try to post, and then i'd tell someone they have to log out of quill and log back in and then it will start working again
#
aaronpk
so, not great, but easy to recover from
#
aaronpk
classic "turn it off and back on again"
#
snarfed
story of my career
#
snarfed
hey aaronpk while you're here, any thoughts on the meta redirect discussion? (https://chat.indieweb.org/dev/2019-04-11#t1555004634348000 and below)
#
Loqi
[snarfed] i have a narrow niche use case for the proxy html+mf2 pages that bridge services like bridgy and bridgy fed render for silo posts
#
aaronpk
oh yeah i vaguely was following that thread
#
aaronpk
i know we've talked about the html meta tags for other HTTP status codes especially 410, but I don't remember any discussion about this for redirects yet
#
GWG
aaronpk, I guess I just need a client to implement them
#
GWG
I will find a partner
#
aaronpk
GWG: there are plenty of other reasons an access token can become invalid other than a natural expiration
#
aaronpk
such as the user explicitly revokes a token
#
GWG
I know
#
aaronpk
so you shouldn't treat this as something that you have to wait for a client to implement
#
GWG
I just think there needs to be a better response on some clients
[stefp] joined the channel
#
GWG
aaronpk, I have it on my list
#
aaronpk
definitely, but you could always lead the charge by implementing that on the server first :)
#
GWG
I already support expiring tokens because I expire authorization codes
#
GWG
it's just a few extra lines
ingoogni, tomasparks, [schmarty], treora, stevestreza, sknebel, snarfed and [kevinmarks] joined the channel
#
[kevinmarks]
RSS is ambiguous about the summary/content distinction - Atom made it explicit. Ideally JSON Feed would follow Atom on this.
#
[kevinmarks]
Both h-feed and Activity Streams make the distinction clear
#
[kevinmarks]
(the original RSS use case had description for links, so was closer to summary than content - the difference was never properly addressed)