#dev 2016-11-04

2016-11-04 UTC
KevinMarks_, KevinMarks, KevinMarks__, miklb and gRegorLove joined the channel
#
www.boffosocko.com edited /2016/LA (+196) "Reminder about daylight savings time changing during camp." (view diff)
miklb, KevinMarks, GWG, KevinMarks_, loicm_, tantek and cweiske joined the channel
#
@MrSimonWood @digisim Not only that, but it doesn’t matter who chooses. LTI, webmentions, micropub etc. what focus of next gen shld be #ngdle #codesign16 (twitter.com/_/status/794474122877276160)
sknebel_, kapowaz, rMdes and Loqi joined the channel
#
sebastianlasse.de edited /2016/Berlin (+11) "/* Participating */ mail confirmed" (view diff)
sknebel joined the channel
#
voxpelli huh, looks like my endpoint has got stuck in an infinite loop of fetching one post each from aaronpk and kylewm
#
aaronpk oh dear
#
voxpelli sorry for inflating any stats :)
#
voxpelli and I'm very happy I have a rate limited fetching system in place ;)
#
voxpelli so – regarding the possibility of two posts mentioning each other and causing an infinite Salmentioning thing? it's real
#
voxpelli or appears to be so at least, still debugging
KartikPrabhu and miklb joined the channel
#
ben_thatmustbeme voxpelli: its totally possible, also be careful of splits and rejoining
#
ben_thatmustbeme in other words, two people reply to a post
#
ben_thatmustbeme another person replys to both of them
#
ben_thatmustbeme forming a diamond
#
ben_thatmustbeme i basically just say "if its in the DB already, assume that url is parsed"
#
ben_thatmustbeme i want to add something to recheck it if its over a certain date
#
voxpelli I thought I had protections as well, I know I at least have it in the data presentation layer
#
voxpelli oh: // TODO: Check if updated prior to _notify(), _ping() and maybe _fetchComments() – else we may get stuck in an inf-loop of pinging!
#
voxpelli Todo-comments only gets you so far
#
aaronpk too bad the computer can't figure out what you mean and just do it itself
#
voxpelli looks like kyle pinged aaronpk when aaronpk's comment was added to kyle's post: https://kylewm.com/2016/05/bridging-the-indieweb-generation-gap
#
Loqi [Kyle Mahan] Bridging the IndieWeb Generation Gap
#
ben_thatmustbeme there is an xkcd about that
#
ben_thatmustbeme trying to find it
#
voxpelli so aaronpk's post contains a comment that is actually his own comment
#
aaronpk it does?
#
aaronpk oh weird, it showed up as a mention
#
aaronpk er rather, one of kyle's comments appears as a mention on my reply
#
aaronpk which is all technically true, just weird
#
voxpelli could be a result of downstream salmentioning
#
aaronpk i may have to rethink how i show mentions
#
aaronpk this mention should not be shown on my page
#
aaronpk (mention = page has a link but is not in-reply-to, etc)
#
cweiske you should only show pages that have the link inside their content
#
ben_thatmustbeme huh... if a post replied to itself, i think that might actually break a lot of things for me
#
aaronpk oh yeah that would solve it
#
ben_thatmustbeme i wouldn't get in to an issue parsing it, but i might hit an infinite loop trying to display it
#
aaronpk i would ignore mentions where my URL is in a child object
#
voxpelli anyone implements a diffing algorithm to know if a mention has actually changed since last seen?
#
aaronpk yeah
#
cweiske my search engine uses e-tag, last-modified and a hash on the content as fallback
#
aaronpk oh wait maybe not. i was thinking of my hub that does.
#
aaronpk my hub won't redeliver a notification unless a hash of the content changed
#
aaronpk is in pubsub land right now
#
voxpelli aaronpk: hash of what content?
#
aaronpk the content at the topic URL
#
voxpelli thinking that eg. a relative timestamp like`This post was written 3 hours, 2 minutes and 32 seconds` would severly break that
#
cweiske it's the last resort
#
cweiske first I use e-tag and last-modified header
#
cweiske the body of the URL
#
aaronpk yes as would any other dynamic content
#
aaronpk Switchboard doesn't use the hash to decide whether to send, just to decide whether *not* to send
#
voxpelli so I think I will do a diff of some more stable parameters and rather miss the mark on some minor details that I don't really care that much about
#
voxpelli hmm, e-tag and last-modified, yeah, that would be a good idea
#
cweiske instead of hashing the whole http response body, one could extract e-content and has that only
#
cweiske s/has/hash/
#
voxpelli diffing to avoid infinite loop should probably rather accidentally break too much than too little
#
aaronpk yeah if the page has mf2 i would recommend hashing the whole parsed json
#
cweiske doesn't loqi do s/// anymore?
#
aaronpk s// didn't make it through the Loqi rewrite
#
cweiske but my s/has/hash/ above?
#
voxpelli implied name would make it hard to hash the entire parsed json I think
#
aaronpk how so?
#
voxpelli "use the textContent of the .h-x for name" -> textContent contains relative date -> boom
#
aaronpk eh that's a pretty edge case
#
voxpelli that someone has a relative date within an h-entry that has no explicit p-name set on it?
#
aaronpk if your implied name contains a relative date then you're doing it wrong and you should fix your markup
#
voxpelli well, I can't fix someone elses infinitely looping markup ;)
#
aaronpk true. code defensively.
#
voxpelli is it possible to know if a p-name was implied or not?
#
aaronpk i don't think so
#
ben_thatmustbeme found it, finally
#
ben_thatmustbeme re: aaronpk saying computers parsing TODO statements http://www.commitstrip.com/en/2016/08/25/a-very-comprehensive-and-precise-spec/
#
aaronpk lol ben_thatmustbeme
#
voxpelli oh, just realized that diffing isn't enough for Salmentions, there just needs to be an explicit loop detection so that a single ping can never result in a post getting fetched more than once
#
voxpelli as Salmention often means that one of the comments has been modified
#
voxpelli and that's the chain I'm currently in need of breaking, because it has probably for a while now tried to find the end of the tree that a ping caused it to start traversing
#
aaronpk cweiske: is http://push-tester.cweiske.de/ still active? i'm not seeing it try to verify my subscription
#
cweiske erm
#
cweiske phubb worker processes were not running
#
aaronpk ah
#
aaronpk fun
#
cweiske had massive server problems in the last days
#
cweiske people are using my push server for themselves
#
aaronpk i saw there was a spam blog that is using all 3 superfeedr, phubb and switchboard!
#
cweiske inters4s?
#
KartikPrabhu first indieweb spam case?
#
cweiske aaronpk, try again
#
aaronpk yeah i think that was the one
#
cweiske workers are running
#
aaronpk hm i'm still not getting the verify request
#
cweiske but tomorrow in berlin I should have time
#
cweiske i'll need to head of to my train soonish and can't check now
#
aaronpk no problem, just wanted something else to test this with
#
aaronpk i'm basically building a tool like push-tester into pubsub.rocks now
#
GWG How many rocks do you have?
#
aaronpk 3
#
cweiske I guess I have to add blacklisting URLs to phubb
#
aaronpk i probably should too
#
cweiske yes
#
aaronpk whoa someone tried to publish a youtube xml file via switchboard
#
cweiske and there is bettsykitchen that I'll blacklist,t oo
#
aaronpk are you going to blacklist that spam one?
#
cweiske angular.jsnews.io wtf
#
aaronpk i have bettyskitchen too
#
aaronpk that looks legit tho
#
cweiske I still don't want to be its host.
#
aaronpk ah
#
aaronpk you might want to add a signup form then
#
aaronpk we must have gotten added to some wordpress plugin
#
aaronpk https://media.aaronpk.com/file-20161104082902-6057.txt
#
KartikPrabhu so many hubs for one site
#
ben.thatmustbe.me edited /2016/MIT2/Guest_List (+116) "/* RSVPs from other sites */" (view diff)
#
ben.thatmustbe.me edited /2016/MIT2/Guest_List (+0) "/* Participants */" (view diff)
#
cweiske .. and before you know you're part of the critical internet infrastructure
#
aaronpk well we're not in pfefferle's default list https://github.com/pubsubhubbub/wordpress-pubsubhubbub/blob/master/pubsubhubbub.php#L139
#
voxpelli To really truly avoid infinite Salmentioning one void have to diff the entire mentions tree after one has traversed it fully to check if something has happened in it I think
#
voxpelli thinks about digging out his old "I have crawled your entire identitity graph so now I can ping you and you you about it"-script
tantek joined the channel
#
tantek.com edited /projects (+249) "restub, make it clear you can do your own project(s) too" (view diff)
#
aaronparecki.com edited /Webmention-faq (-136) "/* What is the canonical URL for the Webmention spec */" (view diff)
#
@iistefka Webmention: A W3C Proposed Recommendation - https://www.w3.org/TR/2016/PR-webmention-20161101 — Front-End Daily (@FrontEndDaily) November 3, 2016 (twitter.com/_/status/794585796024090625)
#
@nexcess Webmentions are the most recent attempt at cracking the nut of inter-blog communication. Let's take a closer look. https://blog.nexcess.net/2016/11/03/webmentions-are-pingbacks-for-the-open-web/ (twitter.com/_/status/794586294223568896)
#
@Gaspertweets nexcess: Webmentions are the most recent attempt at cracking the nut of inter-blog communication. Let's take a clo… https://blog.nexcess.net/2016/11/03/webmentions-are-pingbacks-for-the-open-web/ (twitter.com/_/status/794587560148430848)
#
aaronpk that's a great post!
gRegorLove and tantek joined the channel
#
GWG I still think that the Webmentions plugin needs to be better if it would ever be considered for Core.
KartikPrabhu joined the channel
#
tantek GWG, can we evaluate what "better" means with a Webmention implementation report of what it currently supports?
#
GWG There is a PR
#
miklb as someone who builds stuff with WP, I'm not sure it should be a core feature, just a kick ass plugin
#
aaronpk interesting point
#
aaronpk especially since there are plenty of wordpress sites that don't even have blog posts on them, much less want to show comments
#
miklb exactly
#
miklb but I've always been an advocate for lean and mean with robust plugin api
#
GWG I still think even as a plugin, needs work
#
miklb I don't doubt that. I think if the goal is just be the best plugin it can be, then that helps define goals.
#
miklb s/define goals/define roadmap/
#
GWG I can agree with that
#
GWG That is why I keep coming back to it
#
miklb ?
#
tantek I would add Vouch support as a pre-requisite for Core
#
tantek It has to have a decent (like sizable) security/anti-spam advantage over Pingback
#
tantek especially for wider deployment in WordPress, as that alone would likely tip it into become a spam target
#
gRegorLove aaronpk: Did my mention show up on that post when you're logged in?
#
aaronpk gRegorLove: it did!
#
gRegorLove Sweet!
#
gRegorLove It was an aggressive cache issue. It processed the bearer token correctly, but then still delivered the please authenticate message.
#
aaronpk ah
#
aaronpk voxpelli: doesn't webmention.herokuapp actually reject webmentions where the source URL isn't found?
#
voxpelli aaronpk: absolutely, I still need to check that implementation report
#
voxpelli Got a bit stuck with this infinite loop thing :P
#
aaronpk cool
#
GWG tantek, no one has seriously looked at Vouch for WordPress yet.
#
aaronpk luckily receiving vouches is pretty straightforward
#
GWG I have to reread that section.
#
miklb voxpelli you attending IWC this weekend?
#
voxpelli miklb: unfortunately not :/
#
voxpelli miklb: how's micropub and such going btw?
#
GWG aaronpk, I wouldn't want receive without send. It seems wrong
#
miklb I've been busy with work, but just upgraded my laptop so should be able to get more done in general. Hoping to do a little over the weekend.
#
aaronpk Vouch could always be a separate plugin too
#
GWG aaronpk, there is a hook for sending. Not sure how to separate out receiving
#
GWG So it could be a separate plugin instead of built in.
AngeloGladding joined the channel
#
bischoff.media created /User:Bischoff.media (+194) "Created page with "[[Image:sb.png|right|200px|Sven Bischoff]] = Sven Bischoff = Web Developer from Germany * http://bischoff.media * http://svenbischoff.info * [http://twitter.com/thetalesofimmo ..."" (view diff)
#
bischoff.media uploaded /Special:Log/upload "uploaded "[[File:sb.png]]""
#
bischoff.media created /Template:svenbischoff (+155) "Created page with "<span class="h-card" style="white-space:nowrap">{{sparkline|https://twitter.com/thetalesofimmo/profile_image}} [[User:bischoff.media|Sven Bischoff]]</span>"" (view diff)
#
bischoff.media edited /2016/Berlin (+19) (view diff)
#
bischoff.media edited /User:Bischoff.media (-2) "/* Sven Bischoff */" (view diff)
#
bischoff.media edited /User:Bischoff.media (+3) "/* Sven Bischoff */" (view diff)
#
bischoff.media edited /Template:svenbischoff (+3) (view diff)
#
voxpelli yikes, I found +30k queued up lookups against kylevm's site
#
voxpelli no wonder my fixes didn't have any effect – at the rate it were doing lookups it would have taken around 3 weeks before the queue had been fully processed
KevinMarks joined the channel
#
voxpelli aaronpk: implementation report is updated now
KartikPrabhu and AngeloGl1 joined the channel
#
sebastianlasse.de edited /2016/Berlin (+91) "/* Schedule */" (view diff)
#
@AMKBochnak “We must be willing to let go of the life we planned so as to have the life that is waiting for us.” ― Joseph Campbell #quote #indieauth... (twitter.com/_/status/794645417023643649)
KevinMarks joined the channel
#
KevinMarks Thanks for fixing that - I wasn't sure about some of the error check parts
#
KevinMarks Does kylewm.com have 30k posts?
#
voxpelli KevinMarks: nope, I had a lovely infinite loop within a comment circle that apparently has done 30k iterations around itself :)
#
voxpelli So they all reference the same post, now ensure I never queue up more than one + never iterate over a circle more than once
#
voxpelli (As a Salmention receiver I fetch the full comment trees nowadays)
KevinMarks joined the channel
#
www.svenknebel.de edited /newsletter (+16) "fix double redirect" (view diff)
#
aaronpk voxpelli: thanks!
ChrisAldrich and cweiske joined the channel
#
cweiske aaronpk, https://pubsub.rocks/publisher does not find the hub for http://cweiske.de/tagebuch/
#
cweiske but the http headers contain it
#
cweiske "Your hub and self were not found."
#
cweiske I don't have self though
#
cweiske so at least the "and" is wrong