#dev 2022-03-28

2022-03-28 UTC
[jgmac1106] joined the channel
#
[tantek] I'd like to suggest these instructions instead of what GitHub says to configure a domain for your GitHub static pages, please review: https://indieweb.org/GitHub_Pages#Setup_your_personal_domain
#
[tantek] lol creating new folders in GitHub static pages is so unobvious. you have to create them as part of creating a new file
tracydurnell[d], jacky, cygnoir[d], cybi, [arush], ben_thatmustbeme and feoh joined the channel
#
aaronpk Yeah folders don't exist independently in git
hans63us[d], [arush]1, alex11, reed, diegov, unrelentingtech, kinduff, vladimyr, cambridgeport90[, Vikasheher[m], jamietanna[m], wiki[m], zack[m], denna[m], kloenk, oodani, balupton[d], mro, tetov-irc, cybi, doosboox and jacky joined the channel
#
capjamesg[d] https://www.w3.org/blog/2022/03/a-letter-from-our-ceo-the-web-as-the-ultimate-tool-of-resilience-for-the-world/
#
capjamesg[d] That was a nice read.
#
Loqi [Jeff Jaffe] A letter from our CEO Marking two years since the start of the pandemic, W3C CEO Jeff Jaffe reflects on how the web became the ultimate tool of resilience for the world. The Web as the ultimate tool of resilience for the world It’s hard to draw a t...
jacky, justOkay, mro, cybi, niklasfyi[d] and Seirdy joined the channel
#
[manton] [tantek] [jacky] I took some more time this morning to look at the like MF2 → JSON Feed thing… Granary does produce JSON Feed that seems reasonable from Tantek’s site. For Jacky’s site, there are not any u-like-of attributes. There is e-content, but for some reason it all comes through as a blank string in Granary. I’m not sure why.
Ramon[d] joined the channel
#
[tantek] [manton] interesting. Does it do anything useful with the likes in my homepage mf2?
#
[manton] Nothing special. It just includes “likes this” as the content with a link to the tweet.
#
[manton] But that’s definitely better than an empty string.
#
[manton] I haven’t looked at the Granary source, so I’m not sure where it’s getting tripped up by Jacky’s h-feed. Here’s the MF2 output: http://pin13.net/mf2/?url=https%3A%2F%2Fjacky.wtf%2Fchannel%2Fall
#
[manton] As a test, it would be interesting to add a p-summary and see if Granary picks it up.
#
sknebel that output also doesnt have any like-of properties
#
sknebel and the page source only has "like-of" classes, that is not a mf2 property
[schmarty] and mro joined the channel
#
sknebel so the page markup is the problem
#
[schmarty] this reminds me that there are still a lot of gaps in mf2 tooling
#
sknebel what specifically are you thinking of?
#
[schmarty] well outside my skillset, but i would love to have a version of the web-based mf2 debugging tools that can actually point back into the HTML to show where something is defined
mro_ joined the channel
#
sknebel ah
#
sknebel yeah, I've tried making that once
#
[schmarty] a lot of mf2 "errors" are really tricky because i think the most common are "i am not seeing the thing i expected (because it's mistakenly not mf2 and therefore no parser would it pick up)"
#
sknebel might be worth revisiting
#
sknebel I think I failed first time because I wanted to use the raw source HTML and (html) parsers are not really cooperative with that
#
[schmarty] so on the one hand a "why is it like this" tool that explains where certain things are coming from _seems_ straightforward.
#
sknebel + its a fairly deep partial rewrite of a mf2 parser to expose it well
#
[schmarty] but i don't know how to solve that other class of problems
#
[schmarty] sknebel: yeah it'd require parsers to keep track of where in a document they encountered a particular thing and i don't think any of them remotely do that 😐
#
[manton] I think in this case the lack of u-like-of might be a red herring… It could be improved but the initial problem that we noticed is probably not directly related to the likes but something else with the markup.
#
jacky reads up
#
jacky oh my goodness, I didn't notice that re: my markup D:
#
jacky I'll fix that
chenghiz_ joined the channel
#
jacky okay so I've fixed the lack of`u-` (I really don't know how I missed it)
#
@aswath ↩️ A msg serv can allow anyone to init a session by first auth them against a universal id like Indieauth. The orig serv is not in the picture. #EnThinnai (twitter.com/_/status/1508495012262252545)
mro, P1000[d], barnaby and lanodan joined the channel
#
[tantek] sknebel, which page source? mine or jacky's?
#
sknebel jackys that didt work
#
sknebel *didnt
#
[tantek] [manton] you should see slightly more than "likes this", e.g. "likes @tomcoates’s tweet"
#
[tantek] is the u-like-of making it into the JSON feed?
#
[manton] [tantek] I’m not looking at it right now, but if I remember… “likes this” shows up in the content, and “likes tomcoates’s tweet” shows up in the title and summary fields.
#
[tantek] weird because nowhere do I say "likes this", and my like posts do not have an explicit content property so perhaps it's being auto-generated?
#
[manton] Yeah, possibly Granary is noticing the u-like-of and adding that text itself.
#
[tantek] Is that useful or should we file a bug to tell Granary to use the explicit summary instead when there's no content
#
[tantek] seems weird for the "content" to have less information than the "summary"
justOkay joined the channel
#
[manton] Agreed, it seems weird to me too. I guess some of this is a personal preference… Personally I wouldn’t include a “title” field for liked tweets either.
lanodan joined the channel
#
[tantek] that "title" field is like coming from the p-name, which might be implied by the mf2 parsing
#
[tantek] is likely* coming from
mro joined the channel
#
[KevinMarks] It can make sense to have a title/name if the thing you're liking is a longer piece with a title
#
[KevinMarks] There's some trickiness if link previews happen in the side displaying the like, as that can then lead to awkward doubling
#
[tantek] [KevinMarks] no that is conflation of the title/name of the post vs the thing the post is linking to (h-cite of)
#
[tantek] so absolutely not. that's the kind of sloppiness that caused RSS to be inconsistent
#
[KevinMarks] Well, fair, but Atom:source never quite took off, and so it is a way people use various feeds, so for granary to translate will need some complicated heuristics
#
[tantek] it's not really Atom:source either, which is more of a "repost" like thing
[snarfed] joined the channel
#
[snarfed] thanks for digging into this, all! I haven't followed it deeply, but if you conclude that it's a granary bug or missing feature, feel free to file and I'll look!
jacky joined the channel
#
[tantek] snarfed, no worries! I suppose we don't quite have a "consensus" on whether synthetic content is good/bad (I happen to think it's bad, but that's only my personal opinion / evaluation)
#
[tantek] that is, granary creating content like "likes this", instead of leaving 'content' empty or at least re-using 'summary' if present.
tetov-irc and kimberlyhirsh[d] joined the channel
#
[tantek] Well if we're not going to argue about content vs summary and preserving authored intent vs synthetic values for consuming code convenience, perhaps we can argue about casing conventions: https://twitter.com/pfrazee/status/1508474366111227912
#
@pfrazee Why is it that camelCase looks right in JS but wrong in JSON, while the reverse is true for snake_case (twitter.com/_/status/1508474366111227912)
[aciccarello] joined the channel
#
jacky how do people go about restricting the kind of posts a Micropub client can make?
#
jacky I'm thinking about doing something like `post-type:note` to have my endpoint _only_ report the note post type
#
jacky this is mainly since my indieauth setup is decoupled from my micropub setup so I can't have (currently) some kind of hint into this
#
jacky hmm nothing at /scope
#
jacky I might implement this for my own use
Mr_Imposter[d] joined the channel