#aaronpkI thought the authorship algorithm ended after you match the h-card on that page with the author property of the h-entry
#aaronpkah you are right. I was thinking of the last case of 7
#aaronpkI would add a step which is normalizing URLs, which basically would add "/" to a URL that has no path, since a URL has to have a path in order to do anything with it
#aaronpknote it's not a "trailing slash" rule, it's explicitly only adding a slash if there is no path
#mblaneyoh I'm not sure now... I don't really understand what "author-page" means anyway!
#mblaneywhy should I look on another page when there's a matching h-card on the current one?
#aaronpkthere is potentially richer information on the other page
#aaronpkbut there's no guarantee of that. it might make sense to prioritize using the matching h-card and making it optional to fetch the other page
#aaronpkI would be curious to hear tantek's thoughts on that since he's the one who wrote this up. He tends to have good reasons for these details even if they are not yet captured on the page.
#mblaneyok for now I'll just think about adding slashes then.
#mblaneyhadn't thought about url's needing paths before.
#aaronpkin either case I would still add a slash if there is no path in the URL
#aaronpkthink about the HTTP request involved, if you are trying to fetch "https://bear.im" you make an HTTP GET request to the server at bear.im, you have to request *some* path
#aaronpkso you'd say "GET / HTTP/1.1\r\nHost: bear.im"
#aaronpkoops. i gotta figure out a better way to push updates to my site. it was definitely broken for a minute
#bearyea, that gets me also - i'm gathering quite the collection of small tools that make up my "site"
#aaronpkproblem is I do a `git pull`, but then a bunch of code references things that aren't there until I run `composer install`, and then I still have an issue because database changes aren't applied until I run `migrate`
#aaronpkI guess ruby apps don't have this problem because they don't reload the code from disk until you explicitly restart the app
#aaronpkI wonder how Known deals with in-place updates like this though
#mblaneyalso found your log via new rel=feed discovery changes I've made, so double plus good.
#kylewmaaronpk: known does not use composer (yet) and migrates the db the first time it's accessed
#aaronpkhow does it deal with locking the migration to make sure it doesn't try to migrate it twice
#aaronpkfor example i've got this bot crawling me right now, making 2-3 requests per second, so I don't trust that an in-process migration would finish before the next request came in
#ben_thatmustbeme!tell tantek I have to agree, month names would be clearer. Also, language isn't as much a concern as wikis have a method for alternate language versions of a page, plus translation software will work perfectly well on a month name
#[aaronpk]It also occurred to me that the mechanism for returning a webmention status URL conflicts with the way syndicating via webmention returns the URL of the page that was created
shiflett, lukebrooker, desttinghim, j12t and snarfed joined the channel
#aaronpkkylewm: what are the odds of adding github to silo.pub?
kmahan and raucao joined the channel
#mblaneyhmmm I think bridgy_omit_link has stopped working?
#Loqitantek: ben_thatmustbeme left you a message 2 hours, 22 minutes ago: I have to agree, month names would be clearer. Also, language isn't as much a concern as wikis have a method for alternate language versions of a page, plus translation software will work perfectly well on a month name http://indiewebcamp.com/irc/2016-02-07/line/1454903074408
#tantek! tell ben_thatmustbeme problem with month names is that it's too much work and too easy to get wrong the month name vs. the ISO Date for the <time> elements. and few enough people update /Events anyway so month names = BAD because they make it harder to update.
#tantek!tell ben_thatmustbeme problem with month names is that it's too much work and too easy to get wrong the month name vs. the ISO Date for the <time> elements. and few enough people update /Events anyway so month names = BAD because they make it harder to update.
#KartikPrabhumy homepage does not list "notes" but I use rel=feed to link to an h-feed of my notes, which is consumed by bridgy for POSSE-post-discovery
#KartikPrabhuthat page does not exist should I make one?
#mblaneytantek I've just added support for rel=feed to my reader, specifically to find published h-feeds
[aaronpk] joined the channel
#[aaronpk]The events page is getting to the point where I'm considering setting up a calagator instance that we can use instead
#mblaneythanks to KartikPrabhu for the nudge to support it.
#tantekaaronpk: calagator has the same problem as all other event systems, too much structure getting in the way of being able to informally add random events like people speaking, or uncertain events, or potential events etc.
#[aaronpk]It's actually really not that bad, and it's getting to the point where adding anything to the events page is more work
#tantekbecause we use the wiki to schedule and plan events, the flexibility is important
#KartikPrabhumblaney: My nudge might have been premature as it seems only I publish them and only Bridgy actually consumes them
#[aaronpk]The events page is not used to plan things, it only lists already planned things
#mblaneyno bear supports it too, wouldn't have found his log without it.
#tantekaaronpk: in practice very few people actually help out with events on the site
#tantekaaronpk: not true - we put up rough months when we have an idea for indiewebcamps
#tantekif you think you know how to do events well with structure, great, implement /event posts on your site and show everyone the screenshots of your UI
#[aaronpk]The PDX calagator is a pretty convincing example for me though
#Loqirel-feed is the standard for linking to multiple (potentially alternative) h-feeds from a site's homepage using the code rel="feed" on those links http://indiewebcamp.com/rel-feed
#[aaronpk]Because I want events on my site to exist in the same space as the rest of my posts
#tantekpretty funny that we are sensitive to different things seeming "hard", e.g. aaronpk's point about editing/adding events to wiki is enough for him to not do it, whereas for me, all the git command-line mumbo jumbo is bad enough for me to never use it
#tantekIMO git CLI is far worse than anything on the wiki, events or otherwise.
#tantekKartikPrabhu: nah, with wiki editing, you can easily see existing examples, copy paste, not feel at risk of collossally screwing things up with the slightest typo
#Loqitantek meant to say: kylewm: let me know when git introduces "git undo" to undo the last command you did no matter what it did and then I'll reconsider
#cweiskeaaronpk, the micropub draft only gives an examples for objects that can be created. Shouldn't that be part of the spec? Because in the end, everyone could support micropub and still be incompatible with everyone else
jgarber joined the channel
#borisschapirahello @aaronpk, thanks for the invite
#aaronpkcweiske: shouldn't what be part of the spec?
#Loqiben_thatmustbeme: tantek left you a message 8 hours, 49 minutes ago: problem with month names is that it's too much work and too easy to get wrong the month name vs. the ISO Date for the <time> elements. and few enough people update /Events anyway so month names = BAD because they make it harder to update. http://indiewebcamp.com/irc/2016-02-07/line/1454912283304
#ben_thatmustbeme!tell tantek i'm not sure I agree with you on them being harder to update, I personally think a number is easier to mix up, but thats me. Its really just a problem of people not updating /Events. I wonder if there is some fudgery that can be done with css like time[datetime*="-02-"]::before {content: "February";}
#aaronpkand since each page is marked up with h-event, you could crawl that category page, fetch each URL, and then render a new page showing all the events
#aaronpkthe nice thing is it would be using the canonical event URL page to get the data, so we wouldn't have to also worry about duplicating the data between /events and the event permalink
#aaronpkokay this is very strange. I forgot to put the google analytics code on my new site, so it has not had the tracking code on it from Jan 30 until today
#aaronpkhowever, I still see data during that time period
#aaronpkand not really showing a decrease in traffic either
#tantekCategory:Events could be a good solution yes - especially helpful with being more DRY
#Loqitantek: ben_thatmustbeme left you a message 2 hours, 7 minutes ago: i'm not sure I agree with you on them being harder to update, I personally think a number is easier to mix up, but thats me. Its really just a problem of people not updating /Events. I wonder if there is some fudgery that can be done with css like time[datetime*="-02-"]::before {content: "February";}http://indiewebcamp.com/irc/2016-02-08/line/1454944200048
#ben_thatmustbemeyou assume there is a stage of "translating it back to a word from a number" which i don't think many people do since they go the other way
tvn joined the channel
#tantekno I'm assuming that by having to duplicate month+number you have more chance of error
#tantekwhereas if you *must* get *just* the number right, you'll focus more on that and thus have a better chance of being correct
laprice joined the channel
#ben_thatmustbemeeither way, i think you are overthinking it, adding a word when you create an event is not really a big leap. its just that its not updated,
snarfed, funwhilelost and gRegorLove joined the channel
#tantekI do think templates for the event list items would be helpful
#tantekjust as they were helpful for guest list RSVPs
#tantekthen it would be easier to experiment with different stylings
#gRegorLoveNot really. I contacted the ACM club at Western, though, so hopefully will get some interest.
uranther, benwerd and LaraFZ joined the channel
#snarfedaaronpk: re de-duping, yes please, prefer u-url to source url
#snarfedactually imho the order of precedence should probably be uid, then u-url, then source
benwerd joined the channel
#aaronpkfollow-up question to that then, it would seem I need to restrict that to the same domain. otherwise you could hijack someone's comment on my site by sending a webmention with their URL
#gRegorLovetantek: I think I prefer full month names when the yyyy is not displayed.
#gRegorLoveI could work on a template that displays the full month name just based on yyyy-mm-dd input, though. I think.
#gRegorLoveSupporting other languages would require a separate template in that case, though.
#gRegorLove(Full month names would be hard-coded in the template)
#gRegorLoveYeah, we can use built-in MediaWiki datetime functions, so the event date can always be entered yyyy-mm-dd and displayed in multiple ways, no manual copy/pasta
#aaronpkthe problem i'm having is wirres.net for some reason generates multiple URLs for each post, and one URL pattern shows up in his PuSH feed, and the other is from his webmention sending. both have the same u-url property on the page, but they are different source URLs
#snarfedso you could just consider two wms dupes if the uids and/or u-urls match *and* the source domains match
#aaronpkso his mentions are showing up twice right now
#snarfedi think the proposal above would de-dupe that situation correctly and still prevent hijacking attempts, right?
LaraFZ, tantek, raretrack, KartikPrabhu, j12t, KevinMarks, shiflett and gRegorLove joined the channel
#tantekanyone heard of singleplatform.com? They seem to be a semi-full-service website publishing / updating hosting service specifically for local businesses (example: http://bluehawaiisf.com )
#tantekthey claim to be very easy to setup etc. could be interesting to research in terms of UI for setting up indieweb sites etc.
#desttinghimI'm trying to get my twitter connected to my domain, but it looks like twitter automatically shortens my link on my profile. How have other people gotten around this?
#jonnybarnesI’m trying to write some phpunit tests. if I run just one specific test where I pass example.org to `getAuthorizationEndpoint()` it returns bool(false) as expected. if it gets run in the test suite where that method has been called and returned https://indieauth.com/auth already, then thats what it returns for example.org
#aaronpkit caches stuff in local variables, is that what's happening?
#Schnitzsnarfed1: OK I posted now from Tumblr, Posseing into FB / Twitter via Bridgy using the same webmention a hrefs as before, no change from my side, lets see whether it gets thru, I'll let you know :-)
snarfed, lukebrooker and LaraFZ joined the channel
#jonnybarnesaaronpk: it definitely is your client caching values, if I uncomment the lines in fetchHead and fetchBody that check for self::$_{headers,body} then my tests pass as I would expect
#aaronpkyou could just reset those to null between test runs
friedcell joined the channel
#aaronpkif you want to change the way the library caches things, I would gladly accept a PR :)
#jonnybarnesyeah but these are funtional tests where I have the line `$response = $this->call('GET', url, ['query' => 'params']);`
#aaronpkcool. i would accept that as a PR if you can do that
#jonnybarneshowever this does mean example.org/post1 and example.org/post2 are cached seperately where as you would assume they’d have the same endpoint information
desttinghim joined the channel
#aaronpkthey can have different information though
#aaronpkand since this is for indieauth, it's possible you'd be talking about two different user accounts example.com/user1 example.com/user2
#snarfedhuh. i actually interpreted the idea as more of a planet, ie it would take a set of sites and merge all of their notes automatically into a single feed
#tanteksnarfed, have you seen "planets" that are not *just* article aggregators?
#tantekaaronpk, point is, people now understand an aggregation of public notes as a "public timeline" since Twitter set that expectation
#tantekso if you want to label something that does that in a way that at least *some* people will understand, "public timeline" is MUCH better than "planet"
#aaronpki'm not totally sure I agree with that assessment though. My first thought when I hear "public timeline" is all public things posted to a specific service, not aggregated.
#tantektheir website on their twitter profile aurore2.com redirects to a FB page that says "Sorry, this content isn't available right now / The link you followed may have expired, or the page may only be visible to an audience you're not in." (not logged in)
#tantekHumanML was an early 2000s OASIS attempt at an XML format to describe human communication, previously at humanmarkup.org (long abandoned / reparked domain).
#snarfedSchnitz: re your bridgy publish twitter failure just now, looks like your account is in a funny auth state. if you sign up for twitter again from http://brid.gy/ , then enable publish, that should fix it
#tantekOASIS is an abbreviation for the Organization for the Advancement of Structured Information Standards, a consortium for making web-like standards, similar to W3C, yet more enterprise/business focused in general.
#aaronpkI have decided the next thing I need to do to further all of my projects is Scribe, a service that parses any URL, and stores an archive of what it finds there
#Schnitzsnarfed trying to refresh responses now manually but that doesn't seem to work, maybe no issue tho, I'll try again with responses from a new post... https://brid.gy/twitter/belp_audio#
#tantekaaronpk: does our IRC archives feed get indexed by superfeedr? ;)