#dev 2025-02-26
2025-02-26 UTC
SylvatiCodes joined the channel
# 
trwnh not too jazzed about "web3" myself, but i think it's a bit disingenuous to lump together various visions like semweb ("3.0") and blockchain. the idea of web 2.0 itself is kind of marketing speak to me but a distinct issue with the rush of "user generated content" is that it was too susceptible to corporate capture and herding those "users" onto facebook/twitter/etc. seems like a worthwile problem to solve to me!
# trwnh now idk what the solution looks like but that NIST paper ain't it
# trwnh that's fair, although being a stickler for semantics and terminology i would tend to refer to such things as "the next web" if not for the name collision with a certain news site
# trwnh whatever it's called though, there probably should be an effort to make it easier to have your information not necessarily bound to specific domains or services. i think the idea of having everyone lease their own domain name is a powerful one but it's also got a lot of barriers to widespread adoption (cost, know-how, upkeep). not like there's anything better yet though...
# trwnh what's the NIST got to say about that :p
# trwnh i do hate phone numbers more than domain names, true
# 
[schmarty] we lamenting deeply entrenched systems?
# trwnh nothing worse than a service that requires a phone number. except maybe a service that requires phone numbers but won't accept google voice
# trwnh wrt indieweb my main pain point is that when you change domain names (not transfer registrars) you have to keep the old domain around if you want old links to resolve. otherwise someone can reclaim your old domain name and do whatever they want with it
# trwnh that's a really hard problem to solve though
# trwnh [snarfed] i know someone who has gone through 6 domain names this year alone
# trwnh yes
# trwnh well, it's a bigger issue for people with identity disturbances
# trwnh if you have a consistent identity then yeah, you can probably go your whole life with just one or maybe two domains
[Sophia_wood] joined the channel
# trwnh so it's probably too forward-thinking for most people, who i suspect don't consider too deeply what happens to their site after they die
# trwnh s/this year/in 2024
# trwnh (forgot it's 2025 now)
# trwnh in a more personal sense i would like to have the same content served via multiple domain names but not necessarily mark one of them as "canonical" so much as have that canonical identity be external to the access mechanism (so not an https: identifier). been looking into doi, handle, ark, that kind of stuff. but i think you could do it with relative links against a variable base?
# trwnh starting from a foundation of local-first and resource-oriented, you would then need to route between resources rather than routing between machines
# trwnh basically you would discover "me" somehow, (for you it might be https://trwnh.com/ but for me it might be http://localhost/users/a/ instead) and then you would resolve the relative reference "foo" and it would dereference to "whatever i say is foo"
# trwnh basically using the same principles as petnames (me -> foo) or like how dns subdomains work (.com points you to trwnh.com which can point you to foo.trwnh.com)
# trwnh naturally i want to avoid having to assign every single addressable resource its own fqdn
Luther_Gray and SylvatiCodes joined the channel
GuestZero_, grufwub, alephalpha0 and nemonical joined the channel
bugliker, glacier, [lazcorp], [dave], nemonical, thegreekgeek_ and GuestZero joined the channel
# carrvo Ok, so now my profile page has RSS2.0, Atom1.0, and h-feed, but none work with my eBook. Likely a DNS resolution issue...
# carrvo Not quite sure how to externally test all three renditions but I sure felt like the learning was wasted in the end.
claudinec, jjuran, [Murray], [jeremycherfas], [qubyte], vskate and Luther_Gray joined the channel; rolle left the channel
# 
[KevinMarks] I'm trying to let my excessive collwcriln6od domains expire
# [KevinMarks] *collection of
# 
[jeremycherfas] Putting `html body {filter: invert(1);` into my CSS for `@media (prefers-color-scheme: dark)` is more or less miraculous for dark mode, except that when you simulate dark mode in Firefox, the Inspect Accessibility tool doesn’t seem to make the contrast measurements on the inverted background color, unless I am missing something very obvious.
# [jeremycherfas] Which I probably am.
# [jeremycherfas] Last night’s FrESH was SO inspiring.
ttybitnik and dost joined the channel
vskate joined the channel
# [jeremycherfas] It’s actually quite hard to chck the contrast manually as inspecting in simulated dark mode retains the same color values for each element, even though they display differently. Thanks for the tip about aXe4 dev tools. I will check it out.
[manton] and GWG joined the channel
# trwnh doesnm, for the longest time i didn't even have a phone number. i pretty much exclusively use internet communications my whole life. skype, hangouts, discord. all you need is an email address, the real cornerstone of online identity for now
# trwnh it's just that you can't serve a website from an email address
# trwnh you can't type an email address into a web browser and get anything out of it
# trwnh in something like xmpp their uris have an optional userpart to the authority. http instead did something with basic authentication, which seems backwards to me (it applies to your identity, not the resource's identity)
[aciccarello] and [Sophia_wood] joined the channel
# trwnh i think ftp does the same, and it's equally problematic there. you shouldn't be embedding credentials directly into a uri, the uri should identify the resource not mixing concerns with authentication/authorization (unless it's a capability uri)
# [Joe_Crawford] http can accommodate that https://datatracker.ietf.org/doc/html/rfc3986#section-3.2.1 but there were... downsides.
artlung joined the channel
# trwnh yeah the downsides were bc it was used in a fundamentally incorrect way imo. i dislike the http/https/ftp interpretation and find the xmpp interpretation to make more sense for a more personal or person-centric web
# trwnh ie where dns maps to an ip address, what is desired is instead an authority component that maps to a resource not a machine
GuestZero joined the channel
# trwnh xmpp://a@trwnh.com is specifically my representative account, whereas xmpp://trwnh.com is the server host. but if you type http://a@trwnh.com then you might intend "ask the user a" but you are actually saying "ask trwnh.com, and tell them i am a"
# trwnh so the uri for some reason partially identifies you rather than solely identifying the resource you are trying to address
# trwnh consequently a user is never an authority unless they are given a fqdn
# trwnh which means interfacing with dns records in a rather annoying way
# trwnh compare to using base uri + relative references, where you don't need to do that
# trwnh base uri essentially replaces the authority component here
# trwnh tldr: where you draw the partition line between "resource" and "subresource" matters
# [Joe_Crawford] [tantek]++ thanks for the gentle reminder of what this channel is. we like to solve problems people are experiencing. musing on how people without phone numbers get along in the world is not the best use of this channel. but how uri can map to identify does feel like #indieweb-dev if someone is considering implementing some sorts of auth
# 
[mattl] [doesnm] I don't know of anything I use that needs a phone number for me to login. I do know that people sign up for stuff and lose their email address and cannot request a new password and that's a real problem area. I get a lot of that for http://Libre.fm users who signed up 10+ years ago and now don't use their ISP email address or their old .edu email address.
# trwnh i am mainly asking about the link rot and the "change my identity" bits yeah. more than happy to split off the phone number bits to #indieweb-random instead
# trwnh specifically that if you keep "internal" references consistent, then you have more leeway in what happens "externally", ie how people route to other authorities
bterry, vskate, ezalb, nemonical, Toxicogenic, aciccarello and sebbu2 joined the channel