#dev 2016-11-21

2016-11-21 UTC
tantek and dkm joined the channel
#
tantek.com edited /2016/MIT2 (+40) "/* Sponsors */ Mozilla sponsored coffee and pastires" (view diff)
KevinMarks, KevinMarks_, tantek and cweiske joined the channel
#
cweiske.de edited /Loqi (+113) "/* Commands and Queries */ add !date" (view diff)
voxpelli, plindner, TheGillies, tantek and KartikPrabhu joined the channel
#
miklb.com edited /Habari (+236) "better def" (view diff)
KevinMarks joined the channel
#
@daiyitastic @jasonbrennan YEESSS also this may be relevant: https://indieauth.com/ (twitter.com/_/status/800733613704749056)
tantek, KevinMarks and gRegorLove joined the channel
#
www.boffosocko.com edited /read_later (+232) "silo: Reading.am" (view diff)
KartikPrabhu joined the channel
#
www.boffosocko.com edited /read (+1566) "Chris Aldrich examples" (view diff)
tantek joined the channel
#
loqi.me created /monitoring (+186) "prompted by tantek and dfn added by snarfed" (view diff)
#
snarfed.org edited /monitoring (+793) "expand" (view diff)
#
@salendron Webmentions - if widely available - sounds really interessting! https://twitter.com/programmableweb/status/800777809253830656 (twitter.com/_/status/800781682689376256)
tantek and cweiske joined the channel
#
seblog.nl edited /uid (+6) (view diff)
KevinMarks, pfefferl_, KartikPrabhu, gRegorLove and tantek joined the channel
#
loqi.me created /WebSub (+68) "prompted by tantek and dfn added by tantek" (view diff)
#
snarfed.org created /alerting (+622) "first draft" (view diff)
#
cweiske aaronpk, https://w3c.github.io/pubsub/index.html#discovery says "response from the publisher MUST include at least one Link Header" and a paragraph later "In the absence of HTTP Link headers"
#
cweiske the latter is impossible if the MUST is really a MUST
#
tantek what is alerting?
#
Loqi alerting automatically sends a message (often called a "page") to an app or site's owner or on call ops person when it's down, hitting errors, or otherwise behaving badly https://indieweb.org/alerting
#
tantek what is a pager?
#
Loqi It looks like we don't have a page for "pager" yet. Would you like to create it?
#
tantek what is ops?
#
Loqi It looks like we don't have a page for "ops" yet. Would you like to create it?
#
aaronpk cweiske: that discovery section is about to change to clear that up
#
cweiske to a SHOULD?
#
aaronpk didn't notice that discrepancy before but that's hilarious
#
aaronpk https://github.com/w3c/pubsub/pull/61/files
#
cweiske aaronpk, what does "exact" mean in "The exact URL of the hub(s) designated by the publisher"
#
cweiske ?
#
cweiske does "exact" mean "full" or "absolute"?
#
aaronpk ooh good point
#
aaronpk fully resolved URL as advertised by the publisher
#
aaronpk hm does the spec allow relative URLs? I can't remember
#
cweiske i'm checking web linking rfc
#
cweiske https://tools.ietf.org/html/rfc5988#section-5.1
#
cweiske "If the URI-Reference is relative, parsers
#
cweiske MUST resolve it"
#
cweiske so at least RFC 5988 allows relative ones
#
aaronpk it's worth opening an issue to clarify that
#
aaronpk also if relative URLs are allowed I'll add a test for that to pubsub.rocks
#
cweiske .. or at least remove "exact"
#
cweiske i'll do
#
cweiske https://github.com/w3c/pubsub/issues/66
pfefferl_ joined the channel
#
aaronpk :thumsbup:
#
cweiske following "Appendix A. Notes on Using the Link Header with the HTML4 Format" in discovery opens up a can of worms
#
cweiske - both <a> and <link>
#
cweiske - you must support <base href=".."/>
#
cweiske ..
#
cweiske and it does not define if the first link should be used or the last
#
cweiske or one in the middle
#
cweiske in case I have several <a href=".." rel="hub"> links in my html page
#
tantek "Appendix A" sounds informative not normative :P
#
cweiske websub explicitely refers to that
#
tantek that being said, my PHP link rel parsing library handles that I'm pretty sure
#
tantek (relative discovery)
#
tantek since it passes the webmention tests
#
cweiske aaronpk, is rel=hub body-ok? https://html.spec.whatwg.org/multipage/semantics.html#body-ok
#
aaronpk i would assume so?
#
cweiske since you invent the hub relation, it's not predefined in the html spec
#
cweiske you have to decide
#
tantek right
#
tantek and then note it accordingly in the rel registry
#
tantek what is the rel registry?
#
Loqi It looks like we don't have a page for "rel registry" yet. Would you like to create it?
#
aaronpk any clues from pubsubhubbub 0.3 or 0.4 about whether it would be allowed in the body?
#
tantek rel registry is https://mana.mozilla.org/wiki/display/PR/People+Resources+Home
#
loqi.me created /rel_registry (+110) "prompted by tantek and dfn added by tantek" (view diff)
#
Loqi ok
#
tantek darn copy/paste clipboard fail
#
tantek.com deleted /rel_registry "content was incorrect URL"
#
cweiske aaronpk, 0.4 also only links to appendix a and b of web linking
#
tantek.com created /rel_registry (+123) "rel registry" (view diff)
#
pfefferl_ aaronpk it is not explicitly mentioned in 0.3 and 0.4
#
pfefferl_ s/mentioned/defined
#
cweiske but rfc5988 talks about HTML4, which did not allow <link> in <body>
#
cweiske only <a>
#
cweiske :)
#
tantek rfc5988 is obsolete in lots of ways
#
tantek that being one of them
#
tantek HTML5 supersedes it
#
tantek and the rel-registry does too
#
cweiske I personally don't see the need for <link> in <body> at all
#
tantek cweiske: I haven't found a need for <link> in <body> either
#
pfefferl_ me too
#
tantek FWIW "hub" is already in the rel-registry and supposedly valid on both <a> and <link>
#
cweiske oh
#
tantek per http://microformats.org/wiki/existing-rel-values#HTML5_link_type_extensions
#
Loqi existing rel values
#
pfefferl_ hmm... might result in errors if users can add hubs in comments for example
#
cweiske that'd be an awesome hack
#
tantek lol
#
tantek fortunately first hub wins right?
#
aaronpk nope
#
cweiske that's neither defined
#
aaronpk it's defined
#
aaronpk well, mentioned
#
aaronpk "If more than one URL is specified, it is expected that the publisher pings each of these URLs, so the subscriber may subscribe to one or more of these."
#
cweiske together with fat pings this would mean that I could inject fake news into people's readers when they susbcribe to someone's blog, just by posting a comment with a rel=hub link
#
cweiske the possibilities...
#
tantek ok then
#
tantek it's a good thing we're not depending on any rel values/properties inside of any microformats objects any more then
#
tantek makes it easy to always strip that out
#
cweiske by defining rel=hub NOT body-ok the problem would not exist
#
cweiske hm. but <a rel=hub> still would be allowed
#
tantek right
#
tantek that's the key problem
#
tantek whereas with rel=webmention you only use the first one
#
aaronpk should we restrict to just <link> tags?
#
aaronpk i think it's only expected on <link> tags right now
#
tantek I think that would be reasonable
#
tantek is there an issue open on it?
#
aaronpk i don't think so
#
tantek ok let's open an issue
#
tantek and then also note in the security considerations why implementations MUST NOT support <a rel="hub">
pfefferl_ joined the channel
#
cweiske https://github.com/w3c/pubsub/issues/67
pfefferl_ joined the channel
#
tantek thanks cweiske!
#
loqi.me created /Super_User (+22) "prompted by tantek and dfn added by tantek" (view diff)
#
tantek.com edited /superuser (+276) "/* Silo Examples */ links to Foursquare Super user details, note SU3 password reset" (view diff)