#social 2017-05-15

2017-05-15 UTC
#
sandro Anyway, with my developer hat on, I'd -1 in a second this kind of hostility to fallbacks, since IMO it makes the system just another big silo, maybe even worse than the commercial silos, making innovation impractical. (Since I only have one head, and my W3C hat doesn't like -1's, I'm not sure what I'll actually do.)
#
astronouth7303 I keep thinking of use cases involving publish only systems, eg if GitHub or PlayStation Network had ap implementations
#
tantek well GitHub does have a WebSub (via old PuSH) implementation, and some microformats too IIRC, so we'll see how that plays out
#
tantek GitHub is a WebSub publisher as it were
#
astronouth7303 Did not know#
#
astronouth7303 But those are very certainly cases where there will be specialized extensions
dmitriz joined the channel
#
cwebber2 moin moin
#
cwebber2 <sandro> Meanwhile, it sounds like AP is hostile to fallback
#
cwebber2 ???
#
cwebber2 I'm trying pretty hard to listen to feedback... maybe there are specific things I can do better on? I am open to meta-feedback!
#
jankusanagi_ why would it be hostile to fallback? AFAIK, any new thing you invent, via new verbs, new object types, etc, can use the "content" (or similar name) field as fallback, putting there a plane readable version of your "new stuff"
#
jankusanagi_ at least, that's what you'd do on Pump.io, and I think it applies here too
#
jankusanagi_ ***plain readable xD
#
jankusanagi_ it would also be similar to what XMPP does with the "body" field
#
cwebber2 jankusanagi_: I think, though I'm not sure if it's why sandro is declaring AP "hostile to feedback", that he's frustrated with me suggesting that accepting unknown c2s types is a bad idea?
#
cwebber2 I'm not sure if we've been hostile to feedback in other places
#
cwebber2 maybe I could do a better job on things
#
cwebber2 re: c2s types, I'm not saying an implementation *can't* accept unknown types in c2s, just saying I personally think it's probably not a good idea. The spec doesn't block it, other than the awkwardness around the SHOULD on wrapping non-activity objects in a Create
#
jankusanagi_ cwebber2: he said "hostile to fallback", which I'm taking as "not a typo" :D
#
cwebber2 ohhhhh
#
jankusanagi_ maybe he meant feedback?
#
cwebber2 I misread it!
#
cwebber2 I thought sandro said "hostile to feedback"
#
cwebber2 I'm also pro-fallback on s2s personally :)
#
jankusanagi_ judging by their conversation, "fallback" was the right concept:
#
jankusanagi_ ‎[‎15/5/17 01:53‏] ‎<‎tantek‎>‎ we won't know until we see AP impls try to build in fallback for extensions
#
jankusanagi_ ‎[‎15/5/17 01:54‏] ‎<‎sandro‎>‎ Fallback has to be in the test suite.
#
jankusanagi_ ‎[‎15/5/17 01:54‏] ‎<‎tantek‎>‎ sandro, is what you mean by Fallback in the spec? otherwise we can't put it in the test suite
timbl joined the channel
#
astronouth7303 i think whether servers should accept unknown types in C2S depends on the philosophical role of the provider
#
astronouth7303 if it's like email and providers just act as a relay and repository, then they should absolutely accept any garbage the client throws up
#
astronouth7303 if the server is an active participant and brings value (content) to the process, then maybe they shouldn't?
dmitriz joined the channel
#
ben_thatmustbeme fallback, feedback, thats 'brain only reads the first and last letter' thing has its downsides cwebber2
#
cwebber2 ben_thatmustbeme: arguably, such errors may be part of a "feature" necessary to make progress at all (and so might many "robust" programs we write) :) http://web.mit.edu/jakebeal/www/Publications/NIAI-2008.pdf
#
cwebber2 necessary for humans
#
cwebber2 maybe necessary for *some* classes of programs :)
#
cwebber2 nonetheless I'm glad to be corrected and that my first reading was a misreading :)
timbl joined the channel
#
Loqi Benthatmustbeme made 1 edit to [[Socialwg/AccountDiscovery]] https://www.w3.org/wiki/index.php?diff=102989&oldid=102759
dmitriz joined the channel