elf-pavlik, michielbdejong, mlncn and tantek joined the channel
#tantek.comedited /HTTPS (+74) "emphasize *today* for obtain, don't need any vapor here diluting the effectiveness/actionability of this section. move Let's Encrypt to possible future" (view diff)
#KevinMarks_!tell Tantek Facebook's present continuous "likes" is coherent, but also serves them as they like to imply that liking a brand is ongoing permission for it to post to your feed
#GWGKevinMarks_: You've inspired me to update the wiki
#ben_thatmustKevinMarks_: that would be great. I was just going to try to see what I can do with phonegap though, i don't know what it supports/doesn't support
#ben_thatmustI'll dig in to it KevinMarks_ I may be looking at starting from scratch with phonegap first, hopefully get it working on lots of other platforms too
#Loqitantek: KevinMarks_ left you a message 1 hour, 55 minutes ago: Facebook's present continuous "likes" is coherent, but also serves them as they like to imply that liking a brand is ongoing permission for it to post to your feed
#KevinMarks_Extra fun as I have no twitter client on my phone
#tantekthat actually makes sense as "liking a brand" is essentially a proxy for "following a brand"
#tantekand thus having their posts show up in your "reader" view (AKA news feed) makes sense
#KevinMarks_The equivalent for favourite might be "Kevin favors your post"
#tantekthat's interesting as "favors" implies relative to something else
#KevinMarks_Right, but it is not a superlative like favourite
#tantekRight, Flickr diluted the meaning of "favorite". Or perhaps IE did.
#tantekfavorite in common parlance implies above other choices
#KevinMarks_Well, in flickr it is primarily a plural - your favourite photos
#tantekit's odd that "favorite" became a toggle, as it's "normal" meaning is favorite among a set
#tanteke.g. favorite color, favorite ice cream etc.
#KevinMarks_Which is clearer than "your most liked photos"
#tantekit (favorite people) is a dumb patch of a bad UI anyway (contacts). the right answer is to go straight to people-centric comms as blogged. citation in /comms
#tantekno need for a contacts app when you can arrange your contacts by their faces in home screen pages and folders
#tantekall the mobile "contacts" default app UIs are badly polished 1990s turds
#tantekeven Apple's "Contacts" just looks like bad Lotus Notes or PalmOS from back in the day
#snarfedKevinMarks: just fyi, the twitter app on my phone is working fine. it hasn't logged me out
#mapkycaahh... didn't check this window - glad it's working for you now! (new flickr plugin just captures the nsid from the api, so it's a small change)
Sebastien-L joined the channel
#mapkycaupdated the importer to catch the error - basically it wasn't handling the old style syndication structure very well.
#hugoroydif you like fixing bugs, this one is quite annoying :)
#mapkycaI can see that! I'm sure this is'll be on the todo list somewhere, but the fix may be a migration script... essentially a find and replace on the database. :/
#aaronpkdavidmead: interesting. ownyourgram sends your instagram caption as the "content" field which is meant to be the body of the post, not the title
#aaronpkso it must be Known deciding to save it in the title instead
#aaronpkownyourgram also sends a "category" field with the hashtags in your photos, so Known could be looking for that as well
#aaronpkfor example I have several friends who type really long captions on instagram, way too long for a title
#aaronpkin that case it may make more sense to use the hashtags as the title and the caption as the content
#davidmeadi’m trying to think ahead. the hashtags can be clicked when in the content of a Known post to show just those posts. this setup at the moment destroys this
#tanteklikes will have a very different presentation / interaction than any of my existing post types thus it makes sense that they get a very explicit different post kind internally.
#ben_thatmustbemehmm, need to figure out a way to get my google health data (steps per day) to output some other way than manually
#aaronpkso when I favorited three old tweets from my site, they actually show up way down in my list on twitter
#tantekthat's not as useful, and certainly isn't user-centric (the date order you favorited would be you-centric, which for your profile/favorites makes sense)
#aaronpkI also don't see the date the user favorited a tweet returned in the API
#tantekthe act of favoriting, since Bridgy sees it, then sends it on, and you see it in dated lists e.g. on Adactio's posts, when people faved his POSSE tweet for a post.
#tantektherefore the information does exist, they are simply not storing it
#aaronpkbridgy is reading the list of favorites of a tweet from the HTML page
#gRegor`Yeah, Twitter favorites aren't delivered in the API
#gRegor`as far as I remember what snarfed has said
#aaronpkgRegor`: they are, but they are just a list of the tweets, not a list of "favorite" actions
#tantekhow often is Bridgy checking twitter for favorites?
#tantekin checking adactio's site for faves, I found this
#tantekhttps://adactio.com/notes/8030 - using JS to embed a still image from Instagram - that's unnecessary and sad. I wonder if that's a result of using Instagram's OEmbed endpoint (if it has one).
snarfed, stream7_, KevinMarks and frontwards joined the channel
#tantekKevinMarks: I have not yet explored proleptic newcal but there is no reason they shouldn't work as well as (or better than) proleptic Gregorian.
#friedcellnot saying iso year/week are practical for every day use, but neither is UTC for everyone - it's just a way of storing data in a way that all buckets are the same and every point translates to a single point in what users use
#tantek.comedited /like (+2065) "/* Brainstorming */ add mark up a simple favorite of an indieweb post, an indieweb article, collapsed sequential favorites" (view diff)
#KevinMarksa good heuristic is round-tripping conversions and then testing a long range of them
#tantekit's not just a matter of practical or not, it's a matter of how fragile the encoding, how often it results in cognitive dissonance when inspecting data to verify it, which means eventual corruption is far more likely. in that regard (data fidelity longevity) iso weeks are horrible and should be abandoned.
#tantekaaronpk, others working on "like" posts - I've expanded my design of "like" posts from notification, text only, hypertext, to now sample markup for that hypertext.
#friedcellpostgreSQL exports checks in ddl so where each date fits is obvious + I have everything set up so you can just use the root data table for selects and insert which renders anything "inside" that an implementation detail
#aaronpksnarfed: out of curiosity why don't you have a photo in the IRC logs?
#snarfedaaronpk: i'm being obnoxious and stubborn and want it to pull from my site's h-card :P
#tantekwow that was subtle find. I knew what I meant when I wrote the spec, and it sounds like every implementer interpreted it consistently among themselves, but differently than what I thought I specd!
#Loqikylewm meant to say: huh, does this mean reply-contexts should also be "u-in-reply-to h-cite"? so that the value is the in-reply-to URL?
#tantekkylewm the goal is to make the simple cases simple to publish *and* consume
#tantekyou're right that a more extensive reply context could be more complicated to both publish and consume
#tantekbut let's at least get the simple cases working simply on both sides
#KartikPrabhuwait, so is i have <div class="u-in-reply-to h-cite"> lots of stuff <a class="u-url"></a> </div> then the value of the outermost u-in-reply-to should be the u-url of the embedded h-cite?
#KartikPrabhuthis is not going to be easy to implement
#kylewmso my problem is that now aaronpk has to handle three cases "in-reply-to" is a string, an object with a "url" property, or an object with just a "value"
#aaronpki'm just saying I have never had that happen to me
#tantekthe point is that *any* property can have a h-x-something embedded on it
#aaronpkI think what KartikPrabhu is saying is how do you know if the "value" property is something created intentionally by the author or the automatically implied value which may not be what the author expected to appear there
#tantekso *any* property you are retrieving and expecting a string for, if you don't get a string (i.e. it's an object), you have to go get the "value" from that object
#aaronpkI have had this same thing happen when trying to use the "name" of h-entrys, where the implied name parsing often generates unexpected and ugly results
#tantekthe "value" property is always intentional by use of the p-* u-* etc. class
#tantekaaronpk - that sounds like examples of bad h-entrys that need explicit p-name properties
#tantekaaronpk sounds like perhaps the default wordpress core hentry with broken themes (that lack the rest of the hentry properties) is the source of the problem
#tantekhmm - perhaps mf1 backcompat parsing should not get any implied properties.
#tantek(since no mf1 authors ever expect implied properties to happen)
#voxpelliand ideally there would be some markup one could add to the original site to enable discovery of that other page – perhaps a simple rel-alternate or something
#tanteklost when typing and didn't see it when reviewing
eburcat joined the channel
#tantek.comedited /like (+62) "/* markup for hypertext design */ fix multiple in-stream sequential like posts markup - was intended to be multiple h-entrys, and use p instead of div" (view diff)
#aaronpkthe only problem I see with the new markup is that the second two entries have a name of just "aaronpk's photo" instead of "likes aaronpk's photo"
#aaronpk(also there's a space that looks out of place ( 23:45) instead of (23:45)
#tantekI could use an abbr to make an explicit p-name with "likes" in it
#aaronpki guess I don't really care that much because as a consumer I do my own thing as soon as I see it has a like-of property, and pretty much ditch the rest of the properties
#tantekyeah darn the extra spaces from trying to make markup look easier to read
#aaronpkbut for consumers not looking for "like-of" it would be nice if the name had something that made more sense
#tantekaaonpk - I figured, however it is still good to provide good p-name fallback for any h-entry
#aaronpkso the reason I wanted to see what other people do on their /mentions-feed is I am considering making that page entirely third-person text, the only place on my site where third-person text would appear
KevinMarks_ joined the channel
#tantek.comedited /like (+143) "/* markup for hypertext design */ use explicit p-name for cleaner result in simple h-entry readers, and abbr for expanded standalone p-names for second and later likes in a sequence" (view diff)
#tantekaaronpk updated now with explicit p-name and abbr for explicit "likes " on each h-entry
#tantekit is likely that consumers of the h-entry will display the published time separately anyway, thus including it in the p-name would look bad in those consumers
#voxpelliSo the "like" status is in the relation to a specific post rather than in the post itself :P
#aaronpkI don't think I've seen any "like" posts that are anything but a like
#KevinMarks_Hang on, not that, the faves/retweets on my posts
#kylewmKevinMarks_: aaronpk: well that's a faithful reproduction of how bridgy presents the like ("X favorited this" and "favorited this" appear separately). so maybe an issue that needs to be fixd in bridgy
#aaronpkkylewm: maybe the bridgy favorites should drop the content property
#aaronpkconsumers not looking for "like-of" would then just use the "name" which would say "X favorited this"
#voxpelliKevinMarks_: Yeah, I so need to fix that issue, adding support for likes will get a long way in that
#snarfedif someone is confident they can get the bridgy markup and content right so that it displays nicely in default comment presentations, ie that don't do anything special for likes etc, i'll *happily* merge a PR
#snarfed(markup is very much not my thing, so i'm unlikely to do it myself)
#snarfedvoxpelli: ah, i assumed these examples were more recent. if it's better now then great!
#aaronpkkylewm: snarfed: yeah looking at this again now that tantek wrote up his markup example, i'm pretty sure dropping the "content" would be the correct thing to do
#snarfedaaronpk: hrm. does that assume consumers detect and handle like-of, etc?
#aaronpkif you don't look at like-of you'd just see the "name"
#aaronpka like might mention another page though, like the bridgy notifications of "X liked a post that linked to Y"
#voxpelliis it written anywhere in the wiki that a post can only be a single kind of interaction?
#aaronpkI wasn't making a broad statement like that
#aaronpkjust saying that "like" posts will only be a like, and not a like *plus* some other type
#voxpelliok, so just a single type of interaction (+ possibly mentioning/linking to something else), but possibly be such an interaction for many targets
#colintedfordHm, maybe they got talked out of it since then or have site issues.
#voxpelliI think one could represent a like, reply + reshare as that as just a reshare with a comment
#kylewmyeah I noticed that too; most of acegiak's posts were "liked and reposted" but now just "reposted"
#voxpelliBut it poses the question what interaction to pick if multiple are available. Or one can just do what tantek suggested earlier I guess – and point out that they should fix it.
#Loqibenwerd: tantek left you a message on 12/21 at 8:31pm: would be great to see a brief blog post / transcript of your explanation of indieweb vs federated wiki as tweeted in https://twitter.com/mshook/status/546883742015299584
#aaronpkI don't know how the routing thingy works in Known, but for some reason that get() function is running even when I make a POST request