#dev 2019-04-12

2019-04-12 UTC
[jgmac1106] and [eddie] joined the channel
#
[eddie]
GWG definitely having a server that does something will be extra encouragement for the clients to support it
#
[eddie]
I know expiring tokens is something that’s on my server and client list as well as pkce
#
GWG
[eddie]: I did PKCE one day.
#
GWG
So maybe a hack day project
[tantek], snarfed, gRegorLove, [fluffy], BrianS, [dougbeal] and [manton] joined the channel
#
[manton]
[kevinmarks] I guess I view it as just one more thing that people have to implement. It's simpler for clients to only need to check for whether there is HTML or text content, and not also check for summary or any other fields.
#
[manton]
I've re-read this part of the JSON Feed spec today and think it's pretty clear, even if it doesn't explicitly say not to have blank content with a non-blank summary. This is only confusing because we're trying to map MF2's idea of summary.
[eddie] joined the channel
#
[eddie]
!tell GWG that's awesome to hear it was so easy! I think I'll definitely add it, I just have to finish my AutoAuth stuff first :)
#
Loqi
Ok, I'll tell them that when I see them next
[shaners], snarfed, gRegorLove_, [chrisaldrich], ingoogni, krychu, swentel, gRegorLove__, [tantek], [Rose], jeremych_, [pfefferle], [jgmac1106], swentie and [kevinmarks] joined the channel
#
[kevinmarks]
no, it's also mapping Atom's idea of summary - think about a blog like daring fireball which has a mix of links to external posts with a short summary/comment and articles with a local permalink and full text as a one example. http://www.intertwingly.net/wiki/pie/Rss20AndAtom10Compared#full
[jgmac1106], [Rose], [dougbeal], KartikPrabhu, snarfed, eli_oat, chimo, [eddie], [galexk] and [manton] joined the channel
#
[manton]
[kevinmarks] For a link blog like that, though, I don't see a problem with never using summary. In fact, Daring Fireball's JSON Feed version does not use summary. The "content" is his commentary and a block quote and link.
#
[manton]
Anyway, ignoring Atom, the question from [schmarty]'s feed is really: if you have a "I read this book" post, should there be anything in the content of the post? And I would argue yes.
[schmarty] joined the channel
#
[schmarty]
haha, here we go!
#
[schmarty]
in html+mf2, i would argue there doesn't need to be content unless i have something to say about it beyond "i read this book."
#
[schmarty]
(ownyourswarm would seem to be an agreeing example - there's no content to a checkin unless you type something in. if you type something in, that's the content)
snarfed joined the channel
#
[schmarty]
i feel like this issue comes up again and again but i can't find it being captured well.
#
[schmarty]
e.g. should photo posts include an image tag in the content? i feel like the consensus was "no" and/or "it makes things hard for bridgy" and/or "it depends".
#
sknebel
for mf2 it's clear, but moving to a different format means that has different rules
#
[schmarty]
all that aside, in the context of JSON Feed (or Atom content), the answer changes because they define "content" differently.
#
[manton]
Let's stick with the book or Swarm example and run with it. If you don't type in any text for what you thought about the book or location check-in, then the summary is something about the book and the content is blank. Okay. So what if you do type something in? The content can't just be what you typed because then a client will include that and ignore summary, so you lose all the metadata!
#
[schmarty]
haha sknebel beat me to it.
#
[schmarty]
manton: in the case of an mf2-handling client, i don't think that's true. i believe it should recognize that it doesn't understand the u-read-of attribute and fall back to showing the summary (and content)
#
[manton]
I'm talking about representing that in Atom or JSON Feed, though.
#
[schmarty]
so in an automatic bridging of mf2 to JSON Feed, if my bridge doesn't understand "checkin" or "read-of", i'd want it to combine summary + content as the JSON Feed content.
[jgmac1106] joined the channel
#
[manton]
Okay, I think that makes sense. Basically, "summary" in mf2 is completely different than summary in Atom or JSON Feed, as far as I can tell.
#
[schmarty]
either way i agree the issue is not on the micro.blog side, which is consuming JSON Feed / Atom / RSS as they are spec'd.
#
[manton]
To be clear, Micro.blog does support Atom's summary. It falls back on that if there is no content. But it doesn't do that for JSON Feed. You could argue that Micro.blog should do that, but I think this core issue remains.
#
[schmarty]
haha, yeah. having the same vocabulary mean different things in different contexts is confusing. :}
#
[manton]
Also to your point about photo posts, Micro.blog does add the img tag directly to the content HTML. I think this makes things more consistent everywhere, and avoids problems like in [aaronpk]'s feed where the photos do not show up in an RSS reader.
#
[schmarty]
again that's content (mf2) vs content(_html|text) (JSON Feed vs content (RSS/Atom)
#
[schmarty]
i don't believe a photo post needs to put the img tag inside of e-content.
#
[schmarty]
*in the html+mf2 sense*
#
[manton]
Yeah, I don't think it needs to either. But I choose to. πŸ™‚
#
[schmarty]
but however you generate your other format feeds, it needs to be there.
#
sknebel
it not just "needs to", it means every reader writer has to add heuristics to filter it out again
#
[schmarty]
sknebel: good point! i will amend my statement to say that in the html+mf2 context, the img tag should not be in e-content.
#
[schmarty]
and the questions of how to make sure the image shows up is on the "client" side, whether that client is displaying posts directly or reformatting them as a bridge to other formats.
#
[schmarty]
... looks like i'mma be writing some bridging code, soon, haha πŸ˜…
#
sknebel
how do you generate your feed now?
#
[manton]
This is why I put the img tag in the content, though. There is no bridge code. The same HTML works everywhere. However, I don't want to completely derail this... Let's stick to the main question about summary. πŸ™‚
#
sknebel
[manton]: the only thing that cares about the mf2 "content" property is mf2-aware code. mf2-aware code does not want the photo in the mf2 "content" property
#
[schmarty]
manton: haha i could definitely talk about that for a while (ok so the photo goes in but what about published date? author info?)
#
[schmarty]
i feel like the summary question was kind of settled? imo, micro.blog is doing the right thing according to the JSON Feed spec and my original understanding was incorrect.
[kevinmarks] joined the channel
#
[kevinmarks]
My way of looking at it is that JSONfeed is a regression to RSS ambiguity of description, when everything since has done the name/summary/content split - Atom, h-feed and Activity Streams
#
sknebel
it's fair enough if you just have one notion of "content" and the backend doesn't understand the notion of a photo property it handles (e.g. if photo is done by just adding the img tag to the stored content), but if you explicitly put the value from a photo property into the e-content while rendering/in the template, I'd ask you to reconsider that
#
[manton]
I guess I consider the "regression" a feature. If someone is typing a new blog post, should they really have to know the difference between summary and content?
#
[Rose]
If we don't want to scare new people off, they probably shouldn't _have_ to know it, but it should also be logical to the uncommon observer what each one means?
#
sknebel
I agree that the summary as fallback thing is kind of odd
#
[manton]
Yeah, that makes sense [Rose]. (By the way, I have my Micro.blog "hat" on right now, just speaking for myself... Some of the other folks involved in JSON Feed might disagree with me.)
#
[manton]
Also, wanted to add: I tested [schmarty]'s feed in Reeder's JSON Feed support and it does fall back on summary if content is blank.
#
[manton]
(Possibly because internally in the app it maps to Atom logic, but that's just a guess.)
eli_oat, jackjamieson, [galexk], snarfed and [Rose] joined the channel
#
GWG
[manton]: Wear your Micro.blog hat to the next IWC
#
Loqi
GWG: [eddie] left you a message 14 hours, 5 minutes ago: that's awesome to hear it was so easy! I think I'll definitely add it, I just have to finish my AutoAuth stuff first :)
[schmarty], [tantek], [shaners], [grantcodes], [Rose], swentel, sebsel and [manton] joined the channel
#
[manton]
@GWG Heh. Now I really need an actual hat!
seblog, [kimberlyhirsh] and [eddie] joined the channel
#
[eddie]
[schmarty] I'm very confused. I would never combine summary + content
#
[eddie]
correction, I have never combined summary and content
#
[eddie]
in terms of Indigenous for iOS it checks for a name, then content, then summary fallback
#
[eddie]
So summary fallback only happens if there is no name and no content
#
swentel
I think I do the same
#
swentel
doing an experiment now with timeline per author, it's SO handy :)
#
[eddie]
that's not so say that's not the right thing to do, but I have never seen it explained that way before
#
[eddie]
swentel That's pretty cool
#
Loqi
[sknebel] #21 retrieve feeds individually
#
swentel
now on to testing a global search
#
jacky
anyone got WebSub hub content in their JSON feeds?
#
jacky
looking for an example
#
jacky
what is JSON Feed
#
Loqi
JSON Feed is a feed file in JSON format https://indieweb.org/JSON_Feed
#
jacky
lol talk about a LMGTFY response
#
sknebel
afaik the spec described how to add websub
#
jacky
vaguely
#
jacky
it mentions hubs
#
jacky
and at first, it reads if you just provide a list of URIs
#
jacky
but it seems like a dictionary/map/object with one key being "type" and the other being ??
#
sknebel
"hubs (very optional, array of objects) describes endpoints that can be used to subscribe to real-time notifications from the publisher of this feed. Each object has a type and url, both of which are required."
#
jacky
puts glasses back on
#
jacky
E_USER ;)
#
jacky
thanks sknebel
#
sknebel
it's a bit odd that the later section doesn't even show an example
#
sknebel
or maybe "not ideal", "odd" might be the wrong conotation
wagle joined the channel
#
jacky
so this is a random thought that's come up more when talking to other devs
#
jacky
the use of classnames for microformats has put them off _hard_
#
jacky
I proposed serving mf2+json with the page via a link header and that seems to fly over easier for a lot of them
#
jacky
I might need to get a more 'scientific' poll but I don't know of any MF2 parser that'd (if it finds a <link rel=alternative type=application/mf2+json) use the mf2+json it finds over parsing a whole page
[schmarty] joined the channel
#
[schmarty]
sidefiles--
#
Loqi
sidefiles has -1 karma over the last year
#
jacky
Granted, it also gives those CSR sites a way in
KartikPrabhu, seblog, leg and [grantcodes] joined the channel
#
[grantcodes]
Yeah the classnames thing is not always the easiest for sure. Really depends how you are programming though
jackjamieson joined the channel
#
[grantcodes]
I think the majority of the issues in the WordPress indieweb stuff are related to that. And I'm sure a lot of other cmss would have the same issue
[manton] joined the channel
#
[manton]
jacky: Let me know if you have any questions. There was some discussion early on that JSON Feed's way of thinking about WebSub wasn't strictly compatible with the WebSub specification... I thought it was fine, but the discussion was also a little discouraging so not much has been done with JSON Feed and WebSub, to my knowledge.
#
[manton]
I'd love to get that moving again and dust off Micro.blog's support for WebSub, which is implemented but I don't think widely used.
[tantek] and [eddie] joined the channel
#
[eddie]
[manton] That would be great. I've tried to implement WebSub with my JSON feed that is used with Micro.blog but I'm pretty sure it's not working
#
[eddie]
from what I can tell, I did everything "right"
#
GWG
WordPress does use the sidefile
#
GWG
By necessity
#
GWG
I am hoping for a very indiewebby weekend
KartikPrabhu, [kevinmarks] and chrisaldrich joined the channel