#dev 2022-02-27

2022-02-27 UTC
jacky and ShinyCyril joined the channel
#
barnaby
do we have an algorithm for “given someone’s homepage, how do I find the URL of their main feed (if they don’t have a homepage feed)
#
barnaby
” closequote
#
barnaby
is it just “look for rel=feed”?
#
GWG
barnaby: You are behind there
#
GWG
It changed more recently.
#
GWG
rel=feed is out
#
barnaby
I guess the reference to rel=feed should be removed from https://indieweb.org/feed#How_To_Consume then
gRegor joined the channel
#
aaronpk
I thought rel=Fred was something different than h-feed
#
aaronpk
Omg autocorrect
#
barnaby
hi Fred
#
barnaby
rel=Fred is an extension to FOAF, for marking up links to people with whom your relationship is simply “Fred”
#
barnaby
from GWG’s link, linked h-feed discovery should now be done via rel=alternate links with a text/mf2+html media type
#
GWG
barnaby: Hasn't yet been fully adopted though
#
barnaby
updating https://indieweb.org/feed#How_To_Consume (and noting optional fallback to past solutions) might be a good step towards adoption
#
GWG
barnaby: Not a bad idea. I don't think we did that after the discussion that prompted that
#
barnaby
rel alternate seems like a weird choice for cases where you’re linking to a feed from a non-feed profile page, rather than the intended usage of linking to an alternative representation of the same content with a different content type
#
barnaby
but I guess you discussed that already :D
#
GWG
barnaby: But isn't that also what rel-alternate for RSS does right now?
#
GWG
A page that has a rel-alternate of text/mf2+html pointing to itself would be as you noted
#
barnaby
afaik the intended usage is that a page showing an HTML representation of a feed uses rel=alternate to link to an alternative representation of the same content, but serialised using RSS/ATOM
#
GWG
barnaby: Intended, but not actual usage in practice
#
barnaby
whereas the use case I was talking about, having a homepage which has no feed, and a link to a separate feed page, is not an alternate representation of the same content
#
GWG
barnaby: I agree with you on the intended use of alternate. But is it how it is actually used in the wild?
#
GWG
I was leaning toward rel-feed myself.
#
barnaby
sometimes? often? but not always, because once feed readers implemented rel=alternate based discovery, people started using it outside how it was intended
#
barnaby
publishing h-feed is so different that I don’t see much benefit to piggybacking on that discovery method, personally
#
GWG
Might be worthy of revisiting, being as not everyone has adopted then
#
barnaby
yeah reading through the archived discussion, I’m not entirely convinced
#
GWG
I was just advocating for better h-feed discovery.
#
barnaby
oh I agree with the need for that!
#
GWG
When I first wrote about it, I was contemplating compound rels.
#
jacky
good convo
#
Loqi
jacky: sknebel left you a message 1 day, 13 hours ago: back to your desktop reader work?
#
jacky
b/c I've been working about that
#
jacky
skenbel: that I am (re:desktop work); got inspired by Tootle and Communique
ShinyCyril, KartikPrabhu, justAstache, Seirdy, edburns[d], jacky, balupton[d] and angelo joined the channel
#
[snarfed]
I'm not really on board with the move to rel-alternate yet
#
Loqi
[snarfed] Looked into this a bit. Not easy. There are many times more `rel="alternate"` links in the average page than `rel="feed"` links, and mf2 parsers don't give us their `types`, so we either have to dig it out of the raw HTML ourselves, or fetch (at leas...
cygnoir[d] joined the channel
#
GWG
[snarfed]: So, your first thought is we need to have the mf2 parsers offer types?
#
[snarfed]
I guess?
#
[snarfed]
not proposing a specific solution, just explaining why I can't switch
#
GWG
[snarfed]: I'm just wondering if that is a good thing regardless
gerben joined the channel
#
[snarfed]
"that"?
#
[snarfed]
oh, mf2 parsers offering types. maybe!
#
GWG
Trying to figure out where to propose that.
#
[snarfed]
it would definitely help me switch to rel-alternate
#
GWG
[snarfed]: I'm checking, php-mf2 does return the type in the rel-urls property
#
GWG
So does the python one.
#
GWG
So it is there already
#
[snarfed]
pretty sure it wasn't when I looked, but I can check again
#
[snarfed]
oh interesting, it is there now! wonder if it got added recently. that's great!
#
GWG
I'm going to check if my code uses it, I may have forgotten it was there
BillBennett, angelo and Christian_Olivie joined the channel
#
[KevinMarks]
Type was in the spec for that use case and for media matching
gRegor, gRegorLove_ and mro joined the channel
tetov-irc, [jgmac1106], jacky, bouncepaw[m] and mro joined the channel
#
jacky
that post reminds me a bit of [schmarty]'s post on Webmentions
#
@AlexWilsonV1
↩️ Thanks for the words Rowan - and the gentle reminder to fix webmentions. GitHub Actions & Karaoke really are the best medicine: A good reminder to always channel our inner Kelsier. (that CLI is legitimately so dope - major OBT vibes!)
(twitter.com/_/status/1497950231882829832)
ShinyCyril and mro joined the channel
#
@ariadneconill
↩️ afaik OIDC (and also IndieAuth) are supposed to just pass through unrecognized claims transparently
(twitter.com/_/status/1497955451106996233)
mro joined the channel
#
[KevinMarks]
I suppose that you could map webmentions into a micropub post to create comments, the way bridgy does with blogger comments.
mro joined the channel
#
sknebel
https://typesense.org/ I love the tagline: "Lightning-fast, Open Source Search.No PhD required." ;)
#
Loqi
ok, I added "https://typesense.org/ open-source search service (hosted version available) - https://typesense.org/typesense-vs-algolia-vs-elasticsearch-vs-meilisearch/" to the "See Also" section of /search https://indieweb.org/wiki/index.php?diff=79802&oldid=79613
#
aaronpk
i've been meaning to try that
#
aaronpk
kinda want to replace my elasticsearch instance
ShinyCyril, jarkad, jacky and mro joined the channel
#
capjamesg[d]
aaronpk what do you use elastic search for?
#
aaronpk
the search box on my site
#
capjamesg[d]
Oh that is cool sknebel?
#
jacky
hm the tool https://gimme-a-token.5eb.nl/ doesn't support PKCE :(
#
Loqi
Gimme an Access Token
#
angelo
i too would like to try typesense but i was dissapointed to discover that it doesn't support "phrase search" https://typesense.org/typesense-vs-algolia-vs-elasticsearch-vs-meilisearch/
#
sknebel
thats kinda surprising
#
sknebel
seems on roadmap for next release at least
#
jacky
mellisearch looks like something I'd want to use from that list
#
jacky
the more I look into reply contexts, the less I want to do them in my own site
#
jacky
I have like ~4 cases for content and that's doubled in the event that someone provides a summary
#
jacky
different cases if it's a photo or video
#
jacky
I kinda want to just have a Web component like <reply-context class="h-cite u-url" href="http://magic.com" /> and let mf2 parsers do that
#
jacky
but in browsers, let some remote service do the magic of 'parsing' it so it looks semi-contextually relevant to its surrounding text
#
jacky
hm that would have to be <reply-context class='h-cite"><a href="http://magic.com" class="u-url">Url</a></reply-context> actually since href pulling it specific to <a><area><link>
#
angelo
oh nice, i guess i'll wait for phrase search to land before i give it a whirl. good find sknebel.
ShinyCyril joined the channel
#
Loqi
[bnfd] #272 exact match and phrase query supported?
ShinyCyril, mro, jacky, Silicon[d], corenominal[d], mro_, KartikPrabhu and [tantek] joined the channel
#
[tantek]
Via #indieweb-meta https://desmondrivet.com/2022/02/27/post-type-org - do we have a page somewhere that discusses the whole periodic debate that occurs here about how many (if more than one) feeds or streams one “should” publish and why or why not? I liked the path of analysis in this blog post and we need to capture more like that to help find common ground and patterns
#
Loqi
[Desmond Rivet] A Place For Everything
jacky, tetov-irc and sarahd[d] joined the channel
#
GWG
[tantek]: I'm on a system that publishes a feed for everything.