#indiewebcamp 2015-11-09

2015-11-09 UTC
lukebrooker joined the channel
#
aaronpk ben_thatmustbeme: does <span href="..."> actually owrk?
#
aaronpk seems like you could change that to an <a> if you already know you have to put it in the href
friedcell joined the channel
#
ben_thatmustbeme hah, span href= well it works for the point of parsable mf2
#
ben_thatmustbeme but that is fixed
#
ben_thatmustbeme [shaners]: where is h-array
#
ben_thatmustbeme please file bug reports on github with a way for me to generate it
#
aaronpk i didn't realize the mf2 parsing spec is general enough to find the href attribute on the span! that's interesting
#
ben_thatmustbeme it just sees is u-url and grabs the href, doesn't care what tag it is
#
ben_thatmustbeme works for anything with href, video, audio, etc
#
ben_thatmustbeme well those use src
#
ben_thatmustbeme but you could i suppose
#
[shaners] ben_thatmustbeme: round tripping: mf2 -> jf2 -> mf2
#
[shaners] http://stream.thatmustbe.us/?url=http%3A%2F%2Fstream.thatmustbe.us%2F%3Furl%3Dhttp%253A%252F%252Fveganstraightedge.com%252Fnotes%252F2015%252F10%252F28%252F2%26op%3Dmf2-jf2&op=jf2-mf2
#
aaronpk ben_thatmustbeme: this is already handy for testing what i'm adding to webmention.io
#
ben_thatmustbeme excellent
#
aaronpk although now i kind of want one that is aware of vocab and renders human-friendly versions of the page
#
aaronpk cause what i really want to know is am i providing enough info in this jf2 feed that someone can use it to display it on their site?
cstanhope joined the channel
#
@kylewmahan rhiaro: a drone flies past the window, naturally http://indiewebcamp.com/irc/2015-11-07/line/1446924002898 (twitter.com/_/status/663520390279360512)
tonious joined the channel
#
ben_thatmustbeme aaronpk: that shouldn't be too hard to do
yakker joined the channel
#
kylewm heh, matt mullenweg didn't wait for OCSP to propagate before changing his cert on https://ma.tt/ ... I won't feel bad the next time I do that
#
Loqi kylewm: tantek left you a message 6 hours, 59 minutes ago: is it possible to send a single webmention to Bridgy Publish to get it to publish to all the POSSE destinations (e.g. both FB & Twitter) you've linked to from your permalink? E.g. with 'target=http://brid.gy/publish' ? http://indiewebcamp.com/irc/2015-11-08/line/1447007012870
#
aaronpk ooops
lukebrooker and yakker joined the channel
#
GWG Evening all
#
@InramsT Piwik Is An IndieWeb Alternative To Google Analytics - http:///piwik-is-an-indieweb-alternative-to-google-analytics/ https://twitter.com/InramsT/status/663531327740772353/photo/1 (twitter.com/_/status/663531327740772353)
#
ben_thatmustbeme huh, i never accounted for having more than one type:
#
aaronpk oh yeah we were trying to avoid that with jf2
#
ben_thatmustbeme http://stream.thatmustbe.us/?url=http%3A%2F%2Fveganstraightedge.com%2Fnotes%2F2015%2F10%2F28%2F2&op=mf2-jf2
#
ben_thatmustbeme question is, what do we do if there are two
#
aaronpk are there any use cases for actually publishing more than one?
#
ben_thatmustbeme doesn't matter, there can be
#
aaronpk it does matter
#
aaronpk i was thinking about this today since i'm publishing h-events in my new top-level feed. if someone doesn't recognize how to render an h-event, i want them to do something sensible as a fallback
#
ben_thatmustbeme what do i do well you can say its invalid, but mf2 parser handles it, the question is, what do you do if someone has two
#
ben_thatmustbeme as its valid mf2
yakker joined the channel
#
aaronpk one way to handle the fallback is to make my post both an h-event and h-entry, so that if they don't recognize h-event they'll see the h-entry
#
aaronpk but at that point i can just not specify h-entry and just put a name and content property on the h-event so that the consumer finds those properties ignoring the type
#
aaronpk so what i'm wondering is are there actually any real use cases for specifying more than one type?
#
aaronpk cause if not, jf2 can just use the first mf2 type
#
ben_thatmustbeme we are going at this from different sides, you are assuming it'd probably be better to pick known terms first though, like in this case
#
aaronpk how is that what i said?
#
ben_thatmustbeme woops,two thoughts there
#
ben_thatmustbeme didn't clear line
#
ben_thatmustbeme feeding baby at the same time
#
ben_thatmustbeme s/we are going at this from different sides, you are assuming//
#
aaronpk i'm not sure about that, since that essentially kills the ability to add new types
#
aaronpk well, kills the ability to add new types and still have a fallback
#
ben_thatmustbeme if you just pick first though that can be messy
#
ben_thatmustbeme does mf2 parser order them?
#
aaronpk i was assuming the mf2 parser returns them in the same order as the html
#
aaronpk but i haven't actually verified that now that i think about it
dannyid and [kevinmarks] joined the channel
#
[kevinmarks] That is usually true for lists, but not dicts. PHP preserves dict creation order, python returns them in hash order, go randomises them so you don't accidentally rely on order
#
aaronpk but... we're talking about lists
#
david.shanske.com edited /User:David.shanske.com (-65) "/* Itches */" (view diff)
yakker joined the channel
#
ben_thatmustbeme probably depends on implementation
snarfed joined the channel
#
ben_thatmustbeme aaronpk: the only reason for two classes I can think of is fallback. If that's the case and we want to have only 1 classes we have to take the fallback option always
snarfed joined the channel
#
aaronpk adding a test suite to webmention.io now...
#
aaronpk this is likely what i will be working on during the plane flight back to portland tomorrow
#
GWG Have a nice flight
#
aaronpk thanks!
#
snarfed tests++
#
Loqi tests has 16 karma
[shaners] joined the channel
#
[shaners] aaronpk++ just because
#
GWG snarfed: I'm getting there
#
[shaners] aaronpk++
#
Loqi aaronpk has 1 karma
#
Loqi aaronpk has 2 karma
#
snarfed heh
#
GWG aaronpk: Did you reset?
#
aaronpk O.o
#
[shaners] HA! Karma can’t go 4 digits?
#
snarfed his last ++ did say 1000
lukebrooker joined the channel
#
aaronpk http://meme.loqi.me/4e89ryEY.jpg
#
aaronpk it finally happened
yakker and snarfed joined the channel
#
ben_thatmustbeme huh... now the question is
#
ben_thatmustbeme aaronpk--
#
Loqi aaronpk has 1 karma
#
ben_thatmustbeme aaronpk--
#
ben_thatmustbeme uhhh loqi?
#
ben_thatmustbeme oh yeah, no doubling
j12t joined the channel
#
GWG Continuing my Indieweb work.
#
ben_thatmustbeme great GWG
#
ben_thatmustbeme what are you working on tonight
#
Loqi It looks like we don't have a page for "you working on tonight" yet. Would you like to create it? https://indiewebcamp.com/s/106i
#
ben_thatmustbeme just about to head to bed myself
#
GWG ben_thatmustbeme: After I finished notifications, I was working on storing mf2 for sites I am reacting to
#
GWG I need to figure out how to clean up my storage and display system.
#
ben_thatmustbeme ahh
#
ben_thatmustbeme you got push notifications working?
#
GWG ben_thatmustbeme: Yes. Now much more useful.
#
ben_thatmustbeme awesome
#
GWG My issue now is mf2 storage and display in WordPress
#
ben_thatmustbeme ahh
#
ben_thatmustbeme yeah, working in wordpress can certainly be limiting i'm sure
#
ben_thatmustbeme though i am like passing out right now
#
ben_thatmustbeme so i'm headed to bed
#
GWG ben_thatmustbeme: Night
sparverius, tantek and snarfed joined the channel
#
asheesh (huh, my irssi window here seems to suggest I was notified, but I don't see how, so if anyone needs to ping me, please ping me again)
lukebrooker, Calli, cweiske, sanduhrs, nitot, elima_, detsk, loic_m, petermolnar, ttepasse, KartikPrabhu1, j12t, glennjones, acegiak, elima, ben_thatmustbeme, friedcell, melvster, adactio, fkooman, tantek, tvn, tvn_ and [aaronpk] joined the channel
#
[aaronpk] Goodbye Boston! It was fun!
#
ben_thatmustbeme waves to [aaronpk] on behalf of Boston
ttepasse, nitot, yakker, tantek, melvster, tantek-ipod, j12t, elima and [tantek] joined the channel
#
[tantek] Anyone seen rhiaro?
#
rhiaro hey tantek
#
rhiaro I left you a message in indiechat
#
rhiaro @tantek
#
rhiaro I looked for you in Flour
#
rhiaro I'm now in a meeting until 9 or 9.30 in building 32
#
rhiaro but I can find you to give you a key afterwards
j12t, hs0ucy, squeakytoy, JonathanNeal, tvn, tantek, mlncn and e-lima joined the channel
#
@jgmac1106 @aaronpk @whaity I back up all my tweets to Google Sheets, and publish conversations I want to archive on my @withknown site. #indieweb (twitter.com/_/status/663720395418284033)
tvn and ttepasse joined the channel
#
tantek is re-reading http://werd.io/2015/open-issues-lessons-learned-building-an-open-source-business
elima_, j12t, scoates, friedcell, snarfed, acegiak, yakker, hs0ucy, wolftune, singpolyma and tvn joined the channel
#
@jgmac1106 @redlocal @benwerd Here was my reply: http://quickthoughts.jgregorymcverry.com/2015/benwerd-first-thank-you-to-and-erin-for-building-a #indieweb (twitter.com/_/status/663748904740069376)
singpolyma and tvn_ joined the channel
#
@jgmac1106 @tinokremer @leolaporte Because they are a paying sponsor. He did give @withknown a full episode of TWIG at launch. #indieweb (twitter.com/_/status/663751362237366273)
#
@jgmac1106 @tinokremer Maybe a community based revue sharing #indieweb podcast network is needed to help support and promote tools like @withknown (twitter.com/_/status/663751634787418112)
tantek, tantek-ipod, shiflett and scoates joined the channel
#
@bluesteinwatch RT @jgmac1106: @tinokremer @leolaporte Because they are a paying sponsor. He did give @withknown a full episode of TWIG at launch. #indieweb (twitter.com/_/status/663756182004219904)
#
ben_thatmustbeme tantek, aaronpk was trying to figure if there is a need to support multiple types (multiple h-*) ever? whats the use case? the only one i could come up with is fall back. This came up since jf2 tries to use only one single type, and shaners is using class="h-entry h-as-note"
#
GWG I thought we solved aaronpk's problems
snarfed joined the channel
#
GWG I have to figure out what I want to do with event types
#
GWG I don't implement them currently.
snarfed, benwerd, j12t, mlncn, cleverdevil, Calli and [shaners] joined the channel
#
[shaners] ben_thatmustbeme: I'm moving away from h-as-* classes. I only ever used it for type specific styles. And no one else ever adopted it either.
wolftune and snarfed joined the channel
#
GWG Pfefferle put h-as into the WordPress plugins
#
ben_thatmustbeme ick
[aaronpk] joined the channel
#
[aaronpk] Back on the west coast woo!
#
ben_thatmustbeme awww
#
ben_thatmustbeme :P
[snarfed] joined the channel
#
[snarfed] shaners: bridgy/granary have also used h-as-* forever
#
[snarfed] but kylewm is gradually eradicating then
#
kylewm :D
#
GWG Back in NYC.
#
GWG Pfefferle seems against removing them
#
@WebmontagGi @LukasRosenstock @WebmontagGi #wmgi. Das Indieweb (twitter.com/_/status/663782210944782336)
snarfed joined the channel
#
ben_thatmustbeme huh, wonder the reasoning on that
#
ben_thatmustbeme as i understand it was an activity streams thing, and its been removed from the current as2 spec
#
KartikPrabhu it was mostly tantek using them to conform to activity streams or something. But those never were in mf2
j12t and kevinmarks joined the channel
#
@WebmontagGi Coole websites #indieweb #wmgi https://twitter.com/WebmontagGi/status/663785191798894592/photo/1 (twitter.com/_/status/663785191798894592)
#
@WebmontagGi @LukasRosenstock #indieweb. Das war sehr informativ. (twitter.com/_/status/663785704045084672)
edjw, ttepasse and lewisnyman joined the channel
#
@loschi #indieweb Vortrag von @LukasRosenstock auf #wmgi sehr inspirierend! Danke! (twitter.com/_/status/663791269953601537)
yakker, Calli, cleverdevil-, Tino and [shaners] joined the channel
#
[shaners] Snarfed kylwm hn. I did not know that others were using them. But I'm glad to hear they're going away. :grin:
tvn joined the channel
#
ben_thatmustbeme yeah, activitystreams examples had them, they were never part of the spec
sivoais, tantek, petermolnar and [aaronpk] joined the channel
#
[aaronpk] The only thing I can think of for multiple types is to be able to fall back to something other than an h-entry
#
[aaronpk] But you'd still have to provide enough for an h-entry fallback anyway
#
[aaronpk] So it doesn't actually seem very useful
#
tantek right
#
tantek also, can't believe h-as-* is *still* floating around
#
tantek and getting recent incorporation into WP plugins?!?
#
singpolyma I still have my old <span class="verb">..</span> markup on my site ;)
funwhilelost and hs0ucy joined the channel
#
GWG It wasn't recent
#
ben_thatmustbeme [aaronpk]: the end result is that i think it makes more sense to just pick known types out first, (prefer h-entry over h-blahblahblah)
#
ben_thatmustbeme if we are only providing one class we destroy fallback otherwise
#
GWG I had a WordPress Indieweb thought today
#
ben_thatmustbeme if its well established enough the producer can provide it as h-blahblahblah without fall back, and my implementation will do the same
#
ben_thatmustbeme really needs to think before he speaks, a lot of that is unclear
#
GWG Based on the way I designed Post Kinds, my usage of the post type matches h-entry
#
GWG Which means if I ever do another top level property it needs to be a different post type
petermolnar joined the channel
#
GWG Petermolnar, does that make sense?
#
petermolnar hm, what, where?
#
ben_thatmustbeme huh, aaronpk, tantek, i wonder if it makes sense to figure out what h-* item it should pick first if more than one of the known h-* classes are set
#
petermolnar reading back
#
ben_thatmustbeme start with more specific, ending on h-entry as its most generic, or something like that
#
ben_thatmustbeme not sure it matters as I can't think of instances where it makes sense to use more than one
#
petermolnar what is h-as-* ?
#
Loqi It looks like we don't have a page for "h-as-* " yet. Would you like to create it? https://indiewebcamp.com/s/106j
#
GWG I want to write a something on microformats to WordPress
#
GWG So is the post post type the equivalent of h-entry?
#
GWG Which would mean any other top level properties would need a custom post type
#
petermolnar I'm trying to code the decision algorithm for finding the post type based on that is set for a post instead of setting it manually, but yes, it probably would if they are not the same as you already have
#
petermolnar whereas post type could refer to both
#
petermolnar so I would not add another taxonomy
#
petermolnar or maybe as 'cache' or fallback
#
petermolnar but instead, depending on the content ( meta as well ) decide the post h-
#
GWG What is wrong with taxonomies?
#
Loqi It looks like we don't have a page for "wrong with taxonomies" yet. Would you like to create it? https://indiewebcamp.com/s/106k
#
petermolnar nothing, apart from one more box to tick
#
petermolnar and I'm lazy :)
#
petermolnar so if it's possible to apply the same logic as I myself would do by code, the code is better imho
#
GWG You can do that with a taxonomy and set it in the background
#
petermolnar sure, that's what I'm doing with the post type at the moment
#
petermolnar but I don't have to tick the box
#
petermolnar which is good
#
@voxpelli @simevidas @kevinmarks @Snarfed Easier to just use the ".webmention-error" class to style unloadable images in those cases instead (twitter.com/_/status/663812279792754688)
#
voxpelli the new "Tracking Protection" of Firefox 42 is certainly interesting :P Hard coded black lists that blocks things even when not needed – yay!
#
Loqi yay!
#
GWG Why is a check in an h-entry and not an h-event?
#
voxpelli GWG: it represents an activity, not a happening
#
voxpelli and all activities are h-entries
#
voxpelli as far as I see it
#
voxpelli one can even check-in to an event, no?
#
GWG We were discussing this in regard to trips this weekend
shiflett joined the channel
#
GWG Someone had mentioned that they wanted to check out as well as checkin
#
GWG So that makes a check in something with a possibly defined start and end time
#
GWG Which sounds more like an event
#
voxpelli it could perhaps be represented as both?
gRegorLove joined the channel
#
GWG The microformats wiki defines h-entry as episodic or datestamped content
#
voxpelli like follow/unfollow which can both be represented as the activities of following and unfollowing (probably h-entries) and the actual relation state between A and B which would be rel-contact between h-cards
#
kevinmarks so, this weekend I learned that SVG text and web fonts are really annoying. http://www.kevinmarks.com/htmlversustex.html
#
ben_thatmustbeme kevinmarks: the dangers of the internet :P
#
tantek pops back in to see some post type plumbing discussions ;)
#
tantek ben_thatmustbeme: not sure what you're getting at re: h-entry and fallbacks
#
ben_thatmustbeme so, if jf2 wants to only have 1 "type" (classes from mf2) and mf2's json format allows multiple h-* we need to select 1, we were trying to figure out the reason for every needing more than on in mf2, and really the only thing we can figure is that you provide h-entry as a fallback to h-mycustomthing
#
ben_thatmustbeme if my service to convert mf2 to jf2 provides h-mycustomthing, it breaks the desired fallback the mf2 author included
#
sensiblemn please head over to #knownchat if you want to help me figure out how to get known running in a subdirectory using nginx
#
ben_thatmustbeme but providing only the fallback option (in this case h-entry) it basically handicaps adoption of any new values
#
ben_thatmustbeme thus the issue
#
sensiblemn Loqi: you seem smart. can you come to #knownchat and help me?
#
GWG I am still contemplating h-events as making sense for activities as opposed to actions
#
ben_thatmustbeme i think its better to always provide the fallback though, the resulting jf2 is more likely to be parsed sanely
#
ben_thatmustbeme is that a little clearer tantek?
#
voxpelli GWG: if one wants the check-out to publicly show then it needs to be included in ones feed based on when one checks out – and an event would likely be included in ones feed when it is created
#
tantek ben_thatmustbeme: a little. in practice not a lot of people have been publishing multiple h-* types
#
tantek and in practice some of the more common cases like h-as-* have been largely abandoned
#
voxpelli ben_thatmustbeme: tantek: it's an interesting question though – should it be possible to use new types through some kind of progressive enhancement strategy?
#
tantek voxpelli: we have long thought so, e.g. "h-card h-person" or "h-card h-org"
#
tantek or "h-entry h-review"
#
tantek or "h-product h-listing"
#
voxpelli we had that problem with Activity Streams on Flattr – none of the existing types really suited us, but since Status.net was the only big deployed consumer we had to pick the least wrong type it supported
#
voxpelli and could not define a better type that eg. StatusNet could then later have adopted
#
tantek having to pick the "least wrong type" seems like a good summary critique of post type design assumptions
#
snarfed posttypes--
#
Loqi posttypes has -1 karma
#
snarfed inference++
#
Loqi inference has 2 karma
#
tantek ben_thatmustbeme: why not make type an array like category in the examples when there is more than one type?
#
ben_thatmustbeme "h-entry h-review" huh, so it makes sense to do both of those?
#
tantek that seems like a "simple" way of handling it
#
tantek and then anyone that actually cares about the type can go digging
#
tantek in practice I think consuming code will be written to handle properties not types
#
ben_thatmustbeme we had been trying to avoid that, simpler to consume if you know there is only one type
#
voxpelli +1 on array, in order of preference (because classes in HTML are ordered, right?)
#
tantek nah, simpler to consume if you ignore type and just look at properties
#
voxpelli ben_thatmustbeme: only simpler to consume initially, but since it makes progressive enhancement hard/impossible it will grow messy over time
#
tantek this is getting back to the problem of consuming-code design, where consuming code *thinks* it needs a type to do so something when in practice it actually needs to handle the presence/absence of properties
#
ben_thatmustbeme i disagree, as if you need a type, you end up having to figure out type from properties, which can be complex
#
tantek voxpelli: what ? no not-ordered.
#
tantek neither are rel values
#
tantek they are both unordered sets
#
tantek I always wonder the impression of ordered classes/rel values comes from - nowhere has anyone ever said that - thus it must be something about the eyntax perhaps?
#
ben_thatmustbeme yeah, aaronpk originally suggested just taking the first type, but "first" is actually random since its unordered
#
tantek posttypes--
#
Loqi posttypes has -2 karma
#
voxpelli tantek: hm, so it will always be up to the consumer then to pick which representation is the most fitting?
#
tantek voxpelli: rather, practical consumers will discover that no matter what "type" based code they write, they'll have to handle presence/absence of different properties
sammachin1 and [aaronpk] joined the channel
#
[aaronpk] You can't keep doing posttypes-- when it's built into the microformats spec lol
#
tantek so why not skip all the type-based code assumptions, and *only* implement presence/absence of different properties ?
#
ben_thatmustbeme i was thinking the same
#
tantek basically, developers aren't being lazy enough
#
tantek they're not being long term lazy enough in particular
#
ben_thatmustbeme "long term lazy" heh
#
gRegorLove longtermlazy++
#
Loqi longtermlazy has 1 karma
#
voxpelli +1, it would also show if there's some data that is now inferred from the type that should actually be explicitly stated on its own
#
tantek focusing on post types is a sign of inexperienced (like haven't actually built/shipped solutions that interop with a variety of other implementations) or short term thinking in terms of this kind of stuff
#
[aaronpk] So should I publish my contact info in an h-entry on my home page?
#
tantek frankly, I'm now myself running into confronting even my shortlink single character type codes
#
voxpelli [aaronpk]: in an h-typeless ;)
#
tantek since I'm clearly going down the path of my photo posts being at /t... and NOT at /p... as originally planned
#
tantek voxpelli: that's what h-item is for :P
#
voxpelli oh, handy!
#
tantek lol
#
tantek ben_thatmustbeme: for now for jf2, juss stick all the h-* object types into an array if more than one, just like category
#
tantek and then anyone that complains, challenge them to produce URLs to real world examples
#
rhiaro If you're consuming, and someone gives you a type, and you know what properties to expect with that type, and you even trust the data you're consuming, you're probably *still* going to check for them before doing anything with them anyway, right?
#
voxpelli wonders if any indie-reader would support consuming an h-item of h-item:s like an h-feed of h-entries
#
[aaronpk] Yes
petermolnar joined the channel
#
tantek lol: in practice this is always a false assumption: "you know what properties to expect with that type"
#
singpolyma for now I still distinguish between my "long form posts" and my "social stream activity", but otherwise I barely use types on my site
#
rhiaro so if you're always going to check the properties regardless, you might as well not pay attention to the type, so it might as well not be there
#
tantek distinguishing presentation based on properties is sensible, practical, and the approach many of us are shipping
#
tantek just no need for top level explicit typing
#
rhiaro however if you're super lazy or super trusting or have some kind of super generic consumed-data displaying thing, being given a type might be a useful shortcut. But most consuming code isn't going to be like that I suspect.
#
tantek rhiaro: yes to " you're probably *still* going to check for them before doing anything with them anyway"
#
ben_thatmustbeme #pointless arguement, types make sense for some, not for others
#
tantek that's exactly the point
#
tantek so why not just code that first
#
voxpelli I think the only non h-entry types I have are h-card and h-feed – hard to distinguish between a profile and an article if not using h-card on the profile i guess?
#
tantek ben_thatmustbeme: nah, they make sense in theory until you start interoperating with the wild world of public web publishing ;)
#
[aaronpk] If types really don't make sense then should the microformats spec stop using h-entry vs h-card vs others?
#
singpolyma ben_thatmustbeme++
#
Loqi ben_thatmustbeme has 118 karma
#
ben_thatmustbeme i'm with [aaronpk] on that one honestly
#
tantek aaronpk: good question, or rather, what are the use-cases that drive the use of separate h-card, h-entry, h-event etc.
#
voxpelli right now h-entry is used for what feels like 80% of the use cases anyway, the rest 18% being h-card and 2% being all of the rest
#
tantek like h2vx.com
#
rhiaro having types internally in my system helps me query stuff more easily (because I definitely trust and know what to expect from my own data), but I would probably ignore types coming from external sources and just deal with properties
#
singpolyma [aaronpk]: h-entry vs h-card improve authorability
#
singpolyma IME
#
tantek could we rebuild it and have it "try" to convert every h-*
#
ben_thatmustbeme h-card is very useful for representative h-card, or rather, looking for person represented by a url
#
rhiaro yeah so all my stuff on my site is h-entry and I've been meaning to add something into my mp endpoint that lets me specify a different mf type, but maybe I won't bother
#
tantek types as internal system cache based on property testing could make sense rhiaro
#
ben_thatmustbeme authorship discovery, whatever you want to call it
#
ben_thatmustbeme feels like it just started a big kerfuffle
#
rhiaro yeah, I've been thinking about adding types back in internally only
#
singpolyma ben_thatmustbeme: oh yeah, definitely that
#
ben_thatmustbeme also found an excuse to use kerfuffle
#
tantek ben_thatmustbeme: there's also a possible distinction between generic object types, and post types
#
tantek that is, while post types maybe fairly irrelevant
#
tantek it may be that top level object types have some use-cases
#
ben_thatmustbeme indeed
#
tantek (what microformats has h-* objects for)
#
rhiaro yeah especially h-card with name and h-entry with name
#
tantek (but also why we have so few h-* objects)
#
ben_thatmustbeme although i still contend the "note at a location" vs "checkin with text" are different things, but thats for my internal representation
#
rhiaro but even h-event might not be necessary because you can figure that out from start and end times existing
#
tantek people have over time often wanted to make up their own h-* objects, only because they didn't understand they could re-use an existing h-*
#
tantek rhiaro: events may only have a start time
#
ben_thatmustbeme h-feen is also very useful for feed discovery
#
rhiaro ben_thatmustbeme: http://rhiaro.co.uk/2015/05/checkins
#
tantek and maybe that's enough
#
ben_thatmustbeme knowing ahead of time without pulling a URL that something is a feed
#
rhiaro oh yeah, h-feed++
#
tantek ben_thatmustbeme: right, h-feed is an indication of publisher intent
#
ben_thatmustbeme and in practice (for social) thats really all i see people using (in IWC anyway)
#
tantek but then maybe we can put any h-* as children of an h-feed
#
tantek h-card has the use-case of more robust authorship discovery
#
ben_thatmustbeme well, h-event, though i could see arguements that h-event doesn't really have enough need to be its own top level object
#
voxpelli whatever can be expressed as something else than a type probably should be
#
voxpelli sometimes the type might be the easy solution, but expressing the same value as a property could also be possible
#
voxpelli not using types increases the composability of things
#
rhiaro I maintain that top level objects that are similar enough to other top level objects property-wise yet still need to be displayed differently are the only ones that should have types. I'm not sure how many of those there really are.
#
voxpelli it would be interesting to test and see if one could express them as an h-item through additional properties instead and if that would makes things more or less clear
#
ben_thatmustbeme rhiaro: i don't totally agree a note can't have a "location" as you can post a note about something, but again, this falls to your interpretation of some data, it does not work as a general rule for determining intent of the author of the data
#
kevinmarks adds "summary" and "featured" so http://www.unmung.com/storycard?url=http%3A%2F%2Fkevinmarks.com looks better
#
voxpelli kevinmarks: did you see my tweet btw about possibility to add some CSS to "img.webmention-error" to make failed images look better on your site?
#
ben_thatmustbeme rhiaro: case is point "Rattlesnake (a place a few doors down from our office) is really sketchy, gave my co-worker food poisoning" I am not checking in, but i cite the location for others
#
ben_thatmustbeme i am specifically not checking in, because i am not there
#
tantek ben_thatmustbeme: h-event has the use-case currently of h2vx.com - that is, robust automatic .ics file generation (because calendar programs / sites subscribe to that)
#
ben_thatmustbeme can you determine that based on the properties though?
#
tantek unsure - haven't tried
#
tantek it maybe reasonable to keep going and make h-item the generic thing that you use when you have no need for a specific top level type
#
tantek the irony there is that then that naming lines up with RSS <item>
#
voxpelli ben_thatmustbeme: rhiaro: check-in could be done same way as bookmarking / replies etc though? Add an extra "u-in-reply-to"-style to the geo-tag? "p-check-in-at"?
#
ben_thatmustbeme whats the difference between item and entry any way?
#
ben_thatmustbeme just entry is supposed to be in a feed?
#
tantek what is an h-entry
hs0ucy joined the channel
#
Loqi h-entry is the microformats2 vocabulary for marking up blog posts on web sites https://indiewebcamp.com/h-entry
#
tantek heh
#
ben_thatmustbeme ahh
#
tantek and on the microformats definition, episodic / datestamped content
#
tantek on a page without an h-feed, feed consuming code can imply an h-feed by the presence of h-entrys
#
ben_thatmustbeme voxpelli: p/u-check-in-at makes a bit more of sense to me
#
tantek on a page with an h-feed, feed consuming code can treat all children of the h-feed as part of it
#
ben_thatmustbeme more sense than trying to guess at intent
#
tantek voxpelli: how about checking existing examples of checkin markup that is published?
#
tantek e.g. /checkin#IndieWeb_Examples :)
#
tantek ben_thatmustbeme: always good to check for examples in response to queries about "could be done"
#
tantek as in, "could be done?" should be replied with "how is it done currently?"
#
ben_thatmustbeme voxpelli: might be good to throw that in to a brain-storming
#
tantek brainstorms should consider existing practices
#
ben_thatmustbeme tantek: problem with that philosophy is that is discourages experimentation. "no one else is publishing, so its not needed", but no one is publishing it because there is no option for it
#
voxpelli tantek: only existing example of marking up whether a location is a checkin or not that I see while skimming through is "h-as-checkin" – and well, that's what started this discussion ;)
#
tantek ben_thatmustbeme: hah that's /negative_reasoning
#
ben_thatmustbeme and given that its something i have often cited, and others have had to deliberate on (http://rhiaro.co.uk/2015/05/checkins) i think its something that need experimentation
#
snarfed.org edited /checkin (+92) "/* Ryan Barrett */" (view diff)
#
tantek needs *more* experimentation
#
ben_thatmustbeme yes *more*
#
ben_thatmustbeme i should say the problem is that you need to check existing, non-indieweb usage as well
#
tantek ben_thatmustbeme: that's not a problem, that's good practice :)
#
ben_thatmustbeme well, its somewhat incorrect to just say "look at /checkins#IndieWeb_Examples"
#
rhiaro I thought it was standard to check all usage not just indieweb?
#
voxpelli we're also lacking consumers, publishers we have plenty, but whether the publishes data is ambiguous might not truly surface until we have a wide variety of readers using that content
#
ben_thatmustbeme as thats just reinforcing current indieweb markup, not expressing good practice to look outside of indieweb
#
tantek ben_thatmustbeme: not at all! it's completely correct to say look at indieweb examples - no part of that excludes looking further ;)
#
tantek it's like, go look here first
#
tantek no one said where (not) to look second
#
tantek if anything we strongly encourage looking at silo examples of UX for example
#
voxpelli ideally the other indieweb examples would already have documented the research they based their work on in in the wiki I guess
#
tantek voxpelli: precisely!
#
tantek even if they lack their research, you can and should do better, set a better example for the next person that comes along and sees how/what they should do
#
ben_thatmustbeme indeed
nitot joined the channel
#
ben_thatmustbeme however i doubt everyone does as much research as might be expected. I'll admit to just implementing what other Indieweb sites do.
#
ben_thatmustbeme this is especially true when any section has a "How_to_markup" section
#
ben_thatmustbeme see /comments#How_to_markup
#
ben_thatmustbeme rather claims that the research has already been done and some sort of collective decision has been made
#
ben_thatmustbeme s/ I'll admit to just/ I'll admit to sometimes just /
#
snarfed gRegorLove: fixed the bridgy publish you hit just now. try again?
#
gRegorLove Haha. I was just about to file an issue :)
#
gRegorLove Well, I still could if you want. The published logs show "No log found" for the errors.
petermolnar joined the channel
#
snarfed meh logs are best effort
#
snarfed gRegorLove: ok, this time it's your turn for a fix. :P details in the response body
funwhilelost joined the channel
#
gRegorLove Bridgy publish link is there, not sure why it gave that error. I'll try again.
#
gRegorLove Oh, is this the no-www change recently?
snarfed1 joined the channel
#
snarfed1 oh. drop the www from your link. just brid.gy.
#
snarfed1 i'll make it accept both
#
snarfed1 thought i already had
#
gRegorLove Huh. Yeah, it was accepting the www version for quite a while.
friedcell and cleverdevil joined the channel
#
gRegorLove snarfed: Haha, changed it and now got the error "Couldn't find link to https://www.brid.gy/publish/twitter" (on a separate note)
#
snarfed heh, yeah. fix is almost out
#
gRegorLove I'm in ur bridgy breakin things
#
snarfed k fix is out
bengo joined the channel
#
gRegorLove Despite having an HTML5 player, /Soundcloud apparently still requires Flash, at least on desktop. No audio and a notice "It looks like you have a Flash blocker browser setting or extension. Please enable Flash to hear this track."
#
kevinmarks yep, it's crap. you can switch browser signature to iOS
#
gregorlove.com edited /SoundCloud (+329) "/* Criticism */ Flash still required on desktop" (view diff)
#
gregorlove.com uploaded /Special:Log/upload "uploaded "[[File:soundcloud-flash-required-2015-11-09.png]]""
bengo joined the channel
#
gregorlove.com created /User_Agent (+24) "r" (view diff)
e-lima joined the channel
#
GWG So, now I have a problem.
#
GWG I thought I had a clear solution.
#
GWG All the event type posts could be grouped together, and all the entry type posts could be. Now I have to rethink.
bengo joined the channel
#
GWG I may still do it that way, as it makes sense in terms of them not being in the main feed, but in a separate activity feed.
funwhilelost and friedcell joined the channel
#
gRegorLove GWG: I think a feed of "everything" makes sense, and then separate feeds for each type. Perhaps an option to build hybrid feeds of only selected posts, too. That would give the most flexibility.
#
snarfed a la known
#
GWG gRegorLove: I have an existing system to build in though.
#
gRegorLove What's that?
#
GWG gRegorLove: WordPress
#
gRegorLove WordPress is super flexible for querying posts.
#
snarfed GWG: i think i've mentioned that you should look at https://wordpress.org/plugins/ultimate-category-excluder/ for this?
#
snarfed i use it to do exactly this
#
GWG snarfed: It isn't just that. If I build events as a custom post type in WordPress, I can create a Post UI specifically for activity posts. There's a functional reason.
cleverdevil, bengo, tvn_, shiflett and M-RyanRix joined the channel