marclaportebut I added <link rel="me" href="https://twitter.com/marclaporte" /> <link rel="me" href="https://github.com/marclaporte" /> to my source, as documented
tantek.comedited /Vouch (+292) "prepare to use Bret's cleaned up drawing instead at top of page, move photo of whiteboard to section lower down" (view diff)
marclaporteaaronpk_: thanks. So I tried http://marclaporte.com/Contact and I get 0Â supported and verified providers were found! twitter.com/marclaporte An unknown error occurred with this provider. github.com/marclaporte An unknown error occurred with this provider.
ben_thatmustso i'll just keep resubmitting some small set of 6 digit combos (as many as i can fit in the time before changes) and eventually google auth will happen upon those that i am submitting
marclaporteBut that wasn't enough :-) so I have now pivoted to Tiki Suite: http://suite.tiki.org/Tiki+Suite a FLOSS Server, Web, Mobile and Desktop suite with 80%+ of the features all organizations need
ben_thatmustif a person has your login and password, brute force can still be done on two factor too, its just they have to try as many combinations as possible in the 30 second window they have.
gRegor`GWG: Sure, but ben's point is that indieauth should be doing rate limiting for gootle authenticator login. With the other indieauth options, they're probably already doing that on their end (Twitter, Github)
LoqigRegor` meant to say: GWG: Sure, but ben's point is that indieauth should be doing rate limiting for google authenticator login. With the other indieauth options, they're probably already doing that on their end (Twitter, Github)
ben_thatmustyes, and no, i'm saying any other point you have a large possibility of password inputs, the range is much larger when you have a-zA-Z0-9, forget about if you add other characters, but here you have[0-9]{6} always. it basically makes the time factor in it irrelevent
ben_thatmustthe reason it works so well as 2 factor auth is that it obscures your password correctness on a brute force. so if you try to brute force password, even if you get it right you have to provide a proper TOTP, so (properly implemented) you don't know if you guessed the password correctly or the TOTP failed
ben_thatmustso this got mentioned off the cuff after my demo. But currently (and mainly for proof of concept) I am only pulling my vouch suggentions for my site
marclaportefeels nostagic and hopeful as he replaced <link rel="openid.delegate" href="http://marclaporte.myopenid.com/"/> by <link rel="me" href="https://github.com/marclaporte" /> today...
tantek.comedited /Vouch (+953) "document Protocol Summary based on original IRC with some details since, discussed at IndieWebCamp Cambridge. add FAQ re: Is a vouch saying "you may know me from…"" (view diff)
shanersaaronpk_ Nokogiri in Rubyland provides you similar functionality if you ever find yourself needing to do similar thing for indieauth or whatever.
aaronpk_1Yeah I realized pretty quick that a trivial way to hack that is to make a plain text comment that has the text of something that looks like <a href...
ben_thatmust__, lukebrooker, caseorganic, alexhartley, lukebrooker_, Erkan_Yilmaz, dariusdunlap, adactio, ShaneHudson and KartikPrabhu_ joined the channel
ben_thatmustbemei will still include the list of values that should be set as Reply With source, targer, vouch at least if someone hits the error they can dump the headers and find out what they are missing
ben_thatmustbemei think i have also given up the idea on having my vouch search point published. It would be cool to have that as you could look for vouches from any micropub client, but that also exposes an awful lot, especially since it does curls from the request
aaronpk_and I now have a database cache of all outbound domains in my posts so my vouch approval algorithm went from 3 seconds grep to milliseconds DB query
gRegor`I guess it would cause problems if you accidentally had a domain listed that should have been rel-nofollow, and a spammer could easily create an account on that site.
ben_thatmustbemein fact i'd just do it as a page on my site that just didn't include header info, that way i can test if is_site_owner and display needed remove/block links
gRegor`I might alter the vouch checking to not just be the domain, but anything below a certain path. So that way I could allow a vouch from @t but not @twitterspammer
ben_thatmustbemei need a method to week bad links out already from possible vouch list for myself, I have received referrer entries from t.co and indieauth.com
ben_thatmustbemeugh, I need to basically store regex for any sites i want to black list from vouch use (e.g. indieauth.com) since some i may have to do all subdomains and some not
marclaporte.comcreated /OpenHub (+400) "Created page with "The Black Duck Open Hub (formerly Ohloh.net) is an online community and public directory of free and open source software (FOSS), offering analytics and search services for disco..."" (view diff)
reedstrmcatching up on the weekend irc logs - sat. in Cambridge sounds like it was an awesome meeting! I reaaally need to spend some deep thinking time on how indiewebcamp, peer collaboration, and OER (openstax flavored) all fit together ...
reedstrmre: un-meeting org - sounds like 'start with demos, then schedule' is a good way to get over various common problems w/ 'schedule as very first thing'
ben_thatmustbemeaaronpk_ you might want to ignore homepages too or do something smart to find the permalink, if those end up in my vouches it could pretty much mean plenty of invalid vouches going out
Pierre-O, j12t, loic_m_, danlyke and thierrymarianne1 joined the channel
reedstrmoh, never noticed that diff for 301 vs. 302. What browsers have you tested this against? Doesn't seem to be specified in the relevant specs ...
ben_thatmustbemei think i might fragment my MP client in to a ton of little ones, one simple thing per page, so that on mobile I can bookmark them all and save them in 1 folder
aaronpk_I wonder if it makes sense to send human-redable content so micropub endpoints sort of "fall back" to that if they don't understand the "p3k-food" field that is sent
ben_thatmustbemethat way my webmention code doesn't have to hunt all over, and I should not even need an interface to post a specific vouch, the code can handle it all
ben_thatmustbemei think the only security would be you are no longer protected by the obscurity of it. I personally like the idea for giving myself a place to review the list so i can remove / block entries in ti
ben_thatmustbemeif i am considering some site as a authorized mention source, then its not like i'm opening up a hole by publishing the list, i'm still just as vulnerable to spam from those sources, but It does reveal to others that i have that security issue
ben_thatmustbemeit could get really interesting to start pulling in your autorized sites's authorized list and check their lists for sites you have blocked, then you could automatically notify them of entries you would recommend against
shanersOh yeah. I thought I was gonna have to do some extra work for emoji but by the time I got around to it, I got it for free from the combination of Rails 4 + Ruby 2.1 + Postgres.
reedstrmapropos the Midwest consensus map - what sort of maps would include IL ,but not WI, IA, and MN? Hmm, perhaps they had 'Upper Midwest' as another region.
shanersToday I'm working on this list of autoembeds: tweets, gists, thisismyjam, wikipedia, indiegogo, kickstarter, bandcamp, medium, soundcloud, slideshare, speakerdeck, google maps
ben_thatmustbemepros or allowing follow, that page can be used as vouch URL for anyone on it. If I trust them to post on my site, I am going to vouch for them elsewhere also makes finding if I can use a person as a vouch for me super easy. cons, anyone can send a webmention on my behalf easily, with rel=no-follow its much harder, the person i've linked too should have been able to collect the referals already anyway.
ben_thatmustbemethat page could also just be optional or need not even be complete (hide some connections, etc) its only to make receiving mentions from new people easier
ben_thatmustbemewith following, it becomes a duel purpose list, "here's a list of people I trust to not be spamming me" AND "Here's a list of people I trust to not spam others"
gRegor`If I have linked to twitter.com/t/* and tantek has linked to your ben.thatmustbe.me/*, why shouldn't I trust twitter.com/t/* as a vouch URL for you?
danlykeI've got heuristics which grab feeds from /irc-people and there's a lot of emphasis on human readable there, but trying to automate how to get those feeds from what's there is tough.
ben_thatmustbemegRegor` any time its a silo, I can't trust that they don't turn on side bars or promoted tweets, etc, I trust tantek not to post bad content, i don't trust twitter
shanersthat said. i agree with what ben_thatmustbeme just said: recommend it, but don't assume it. infer it if .h-entry(s) are present without an .h-feed
gRegor`I only have one h-feed so far, for my notes. It's linked from my <head> with rel=feed, so in that instance I think the h-feed makes sense as a "requirement"
LoqiA card in the context of the indieweb are small information summaries—often summaries of specific, external web pages http://indiewebcamp.com/cards
LoqiAn autolink is a hyperlink that was automatically added to some text to link that text to an obvious or useful destination, e.g http://indiewebcamp.com/autolink
snarfedapologies, late to the conversation…but re implied vs explicit h-feed, bridgy's original post discovery (using syndication links) currently expects it to be explicit
gRegor`@joeld's original idea was a rel-vouch to a text file of URLs, but it could be to HTML. I don't think it was intended for individual external links.