#dev 2020-03-13
2020-03-13 UTC
[Michael_Beckwit joined the channel
# 
[Michael_Beckwit if it wouldn’t ruin 20+ years of using “tw2113” as a handle, I’d consider “tdubs” at this point
# [LewisCowles] Are they not such a limited resource that it's trivial to count the importance of one anyway?
[LewisCowles] joined the channel
[chrisaldrich], [schmarty], [tantek], geoffo, gRegorLove_, nickodd and [g33kcentric] joined the channel
# [g33kcentric] Ive finally decided to knock having subdomains for auth, micropub etc on the head and just stick the URLs on the end of my domain, just so much easier all round, sometimes im too annoying for my own good :D
[Rose] and gRegorLove joined the channel
cweiske, loicm, Guest2312, [jeremycherfas], nickodd, hs0ucy, [jgmac1106], petermolnar, garrettw, ngbgn and [AlisonW] joined the channel
# wink and I think that's a purely academic difference. I don't see any human regularly typing in any non-obvious parts of a path (which may be there or not). Browsers at some point started looking for favicon and robots, despite them being not for humans. If you link to them the path doesn't matter, and they're not *discoverable* anyway. So by inventing new fixed-urls "for humans" all I see is having an emerging standard pushed on the website owners
# wink in a way that I've not seen any W3C or other standard doing it.
# wink if your pastime is trying out /humans.txt then of course you're right
[LewisCowles] joined the channel
# [LewisCowles] alisonw++ I like the idea of hiding it in yet another DNS TXT record. I actually feel like the clacks overhead I added to my own site should be DNS instead of HTTP header
# [LewisCowles] I'll concede humans.txt and /.well-known/* are better than per-request HTTP headers
# [LewisCowles] Especially when the HTTP headers are often not contextual
# [LewisCowles] It's very hard to experiment with a lot of headers though because some are for site-security, which is a granular, but AWFUL model
# [LewisCowles] It seems it's an alternative to having a WSDL / Swagger JSON/YAML describing all endpoints with patterns and having named security groups (I'd consider future)
# [LewisCowles] If you think about it that could eliminate a lot of headers going back and forth, so up to 8KB each way for a reasonably locked-down system
# [LewisCowles] One other lovely thing to have would be content aware declarative browser effects, such as horizonatal, vertical diagonal bottom-left to top-right and opposite scanlines, which detect K value and adjust between highlight, lowlight and mid-tone (perhaps hints for midtones) so that people can pack in smaller images, with noise effects like older hardware used to use to make graphics that were meh, pop. This is what I was looking for when
# [LewisCowles] remembering every-layout.dev
# [LewisCowles] K being the amount of black in a projective light system, or the absence of red, green, blue
[jeremycherfas], KartikPrabhu, [tantek], [Sadik_Shahadu], nickodd, [Michael_Beckwit, [KevinMarks], [Jeff_Hawkins], [Rose], leg, gRegorLove, [Joe_Crawford], superkuh, [schmarty], jamietanna, [yo], gRegorLove_, [calumryan], [snarfed], neuralsplash, [jgmac1106], djmoch and [AlisonW] joined the channel; nickodd left the channel
# wink now, yes
# wink ok, that's new to me, I only knew it in /
# wink but I don't see any of the 2 points refuting my point in any way
# wink the favicon is actually making it stronger. because it is for machines AND configurable
# wink if there was <link rel="humans.txt" ...
gRegorLove, KartikPrabhu, loicm and [AlisonW] joined the channel