#dev 2021-09-17

2021-09-17 UTC
[tw2113_Slack_], [aciccarello] and shoesNsocks1 joined the channel
#
GWG Under the 2020 Spec Update of IndieAuth, is response_type now mandatory instead of optional?
#
GWG It looks like the word optional was removed.
hendursa1 joined the channel
#
aaronpk that would make sense sincei t's required in oauth
#
GWG aaronpk: I was just looking. I think I left it as optional for backward compatibility. It was optional in the original version, because its absence assumed response_type=id
#
GWG And the only post request the authorization endpoint accepts is grant_type="authorization_code'
maxwelljoslyn[d] joined the channel
#
astralbijection[ how many times are webmentions supposed to be sent?
#
GWG astralbijection[: That depends. If they are updated, more than once
#
astralbijection[ or i guess, can you expect senders to send more than 1 webmention?
#
astralbijection[ GWG: ah, okay
#
GWG If they are never updated, only once.
#
astralbijection[ should they be re-sent if there was a failure?
#
GWG astralbijection[: What sort of failure?
#
astralbijection[ like, the server gives a 500 because i wrote my receiver poorly lol
#
GWG It's an implementation detail.
#
astralbijection[ okay
KartikPrabhu joined the channel
#
GWG I know if I have a 500 on sending, I queue it to retry a certain number of times before giving up
#
GWG Remember, that would mean that the site was advertising an endpoint, so I would have to assume they wanted me to send a webmention
#
[snarfed] https://indieweb.org/Webmention-brainstorming#retrying.2Ferror_handling
sebbu joined the channel
#
GWG Trying to figure out refresh token logic in my system, namely how to tie the refresh token to the access token
#
capjamesg[d] astralbijection[ I think senders might send a second webmention to update.
#
capjamesg[d] But the receiving endpoint would have to support updating.
hendursaga joined the channel
#
sknebel more search engines: https://search.marginalia.nu/ "The search engine calculates a score that aggressively favors text-heavy websites, and punishes those that have too many modern web design features.
#
sknebel " - some more commentary from the creator in https://news.ycombinator.com/item?id=28550764
#
Loqi [Funes-] A search engine that favors text-heavy sites and punishes modern web design
#
capjamesg[d] 😄
#
capjamesg[d] I have seen that search engine actually.
#
capjamesg[d] "Good comparison. Reminds me of an analogy I like to make of today's web, which is it feels like browsing through a magazine store — full of top 10s, shallow wow-factoids, and baity material. I genuinely believe terrible results like this are making society dumber."
#
capjamesg[d] I don't agree with the second sentence but I do the first.
#
capjamesg[d] Love this: https://news.ycombinator.com/item?id=28554483
#
Loqi [slim] This is a search engine indexing the internet on a mariadb database hosted on consumer hardware maintained by a single person as a hobby and it does not suffer from HN hug of death
#
capjamesg[d] With concurrency and a long list of websites, you can do some terrific things with a search indexing algorithm.
#
capjamesg[d] IndieWeb Search has indexed at least 200,000 documents this week (but not in the prod version yet) with a DigitalOcean server.
jjuran joined the channel
#
capjamesg[d] Does Elasticsearch have greater memory requirements at scale?
#
capjamesg[d] I assume so, right?
tetov-irc joined the channel
#
[KevinMarks] scale usually means various things going from RAM to disk, with the consequent ~1000x performance loss. initially all your data is effectively in RAM because of caching, then it is served from disk. Later your indexes no longer fit in RAM and then things really slow down.
#
@samclulow more books being read this year (this is just an excuse post to see if i can get webmentions to work) https://smcllw.me/bookshelf/ (twitter.com/_/status/1438869497482563602)
shoesNsocks joined the channel
#
GWG aaronpk: I was going to update those IndieAuth PRs based on the feedback so far. Do you have any?
#
capjamesg[d] [KevinMarks] I have written a quick program that extracts links from pages in the search index with the long-term goal of adding outgoing links as a ranking factor. But I'm not sure how much disk space I'll have left 😄
#
capjamesg[d] 53,000 links in and over 1.5 million links have been identified.
#
capjamesg[d] (internal and external)
#
capjamesg[d] Doing all of these calculations has made me appreciate how fast modern computers really are.
IWSlackGateway and shoesNsocks joined the channel
#
capjamesg[d] A good read from the creator of the Marginalia search engine: https://search.marginalia.nu/notes.html
#
capjamesg[d] “The HTML standard is not a standard. A major american university uses <title>-tags for its navigational links. It's a technological marvel how coherently web browsers deal with the completely incoherent web they browse.”
[schmarty], shoesNsocks1 and tetov-irc joined the channel