#dev 2018-11-07

2018-11-07 UTC
mblaney, [jgmac1106], deathrow1 and [eddie] joined the channel
#
Loqi
[Charlie Owen] Dear Developer, The Web Isn't About You
#
[jgmac1106]
why is this happening? Do I need to make my h-card representative because the bookmarks on the page have an h-card?
#
[jgmac1106]
nvm, think it;s just here the cited author h-card gets picked up...if I remember correctly granary? favors the h-cite
snarfed joined the channel
#
gRegorLove
carrying over some indiebookclub talk here from #microformats
#
gRegorLove
Grant has a request in for IBC to allow editing posts, so that's on my radar.
#
gRegorLove
In #microformats we were discussing possibly switching `p-read-status` with `p-category`
#
gRegorLove
If I do that, I think edit posts would check both the p-read-status and the p-category to select the "status" in the drop-down...
#
GWG
I am still confused
#
gRegorLove
So if you edited on IBC, changing from to-read to reading, the mp request would send the new category, not the old one.
#
gRegorLove
[eddie], [jgmac1106] Does that make sense?^
#
GWG
But I am jetlagged
[aaronpk] joined the channel
#
gRegorLove
Personally I don't think I plan to use edits in IBC much. I like having separate posts for when I added to-read, started reading, finished reading.
#
[jgmac1106]
yeah part of the reason I was thinking them as a webmention to the original post
#
gRegorLove
I've experimented with that for reading progress, each one in reply to the "started reading" post.
#
gRegorLove
Not sure about it in the mp client, though.
#
[jgmac1106]
but in IWBC could you theoretically have a dropdown menu that could trigger a micropub post for read-to reading and finished that would in a reply-to the original post?
snarfed and snarfed1 joined the channel
#
[eddie]
gRegorLove ohhh interesting. Yeah I would probably do separate posts, I don’t think I would ever edit
#
[eddie]
But makes sense for people that want to do that
#
gRegorLove
[eddie], out of curiosity, are you doing anything with read-status on your site internally, to organize posts?
#
[eddie]
Not yet. Without a database cache, I couldn’t do anything. But now that I have a database cache set up this week, I’ll be able to start doing that
#
gRegorLove
One reason I'm liking the thought of changing to category is it solves the problem of whether to use the x- prefix in the mp requests :)
#
[eddie]
Yeah, I think it’s a pretty elegant solution
#
[eddie]
I’ll be able to watch for the special “categories” and when going from oldest to newest, update the database cache with the most recent book status
#
gRegorLove
Ok, just checking. I don't want to make life difficult, heh. There's still the read-of so people can identify read posts though
#
[eddie]
So that’ll allow me to do separate posts but in the database I’ll be able to query all posts with “to-read@s
#
[eddie]
Yeah, I think that’s the key
#
[eddie]
Having to-read
#
[eddie]
Haha read-of is the key
#
[eddie]
For Post Type Discovery
#
gRegorLove
[jgmac1106], possibly. I'll think about it some more. I want it to be a clear UX and not surprise anyone. Maybe a setting.
snarfed and renem joined the channel
#
@tomcritchlowBOT
V curious about webmentions and the insanity of….
(twtr.io/1kBM6m_o_G_)
cweiske, [pfefferle], [Vincent], swentel, mblaney, GWG, [chrisburnell], [jgmac1106], [schmarty] and [grantcodes] joined the channel
#
[grantcodes]
!tell gRegorLove I've not used IBC extensively yet myself, but the way I would imagine updates working would be practically the same as a new post, but if the user selects they have finished or got to page you could show an extra box to paste a post url to update.
#
Loqi
Ok, I'll tell them that when I see them next
#
[grantcodes]
!ell gRegorLove also seems I can't login anymore. The auth redirect has a double slash in it which I'm sure doesn't help
#
[grantcodes]
!tell gRegorLove also seems I can't login anymore. The auth redirect has a double slash in it which I'm sure doesn't help
#
Loqi
Ok, I'll tell them that when I see them next
[jgmac1106] joined the channel
#
[jgmac1106]
@swentel how do apps access the finerprint reader on Androi...be really cool if I could login with a fingerprint
#
swentel
[jgmac1106], well it's a permission that can be asked by the app and then implementing some methods to do the verification
#
swentel
it's relatively simple
#
[jgmac1106]
...but with indielogin it would be my website that would have to access the fingerprint reader correct?
[Rose] joined the channel
#
[Rose]
Aaron has a post somewhere about using his phone to log into his website
#
sknebel
yes, his IndieAuth endpoint can use some authenticator app on his phone as a second factor
#
sknebel
a client app could protect a token it already obtained behind a fingerprint check, but can't play that role for the actual login to the IndieAuth server
#
sknebel
hm, another interesting point for private (or worse, partially-private (do we have a good word for this?)) feeds is WebSub. I think it supports authentication in principle?
[eddie] joined the channel
#
[eddie]
sknebel: partially-private, could be referred to either as protected or audience based
snarfed and [schmarty] joined the channel
#
aaronpk
sknebel: PubSubHubbub has historically been viewed as being unable to support authenticated fetching, which is why we keep seeing alternative methods of authenticated delivery pop up
#
aaronpk
I think some of that is due to the assumption of a hub doing the "fan-out" thing where it fetches a feed once and delivers the same content to all its subscribers
#
sknebel
yeah, for "mixed" resources it's kind of iffy
#
aaronpk
if we think of a feed not just as a URL but as a URL plus some authentication, then I think the rest of the WebSub model still falls into place nicely
#
sknebel
yeah. it's a bit tricky when and how to authenticate, it's not as obvious as I thought it'd be at first
[tmiller] joined the channel
#
sknebel
I think a better first step would be to write down what we have in a more formal shape, with request examples and such
#
aaronpk
yep and maybe get another implementation or two :)
#
aaronpk
since that private webmention spec is pretty well documented and has a couple implementations but nobody else seems very excited about it
#
sknebel
I can turn my code in a test server
#
sknebel
so everyone can log in and see a "private post" with their log in URL
#
sknebel
so people making the fetch side can see something immediatly
#
aaronpk
that sounds like a good start
#
sknebel
Zegnat had some problem with the indieweb client php thing that for some reason didn't like my 401 pages, but worked for him locally...
#
aaronpk
I still need to review everything again and see what it'll take to add it to my site
#
aaronpk
since I already support private posts addressed to specific people
#
aaronpk
(not feed views, just permalinks)
#
sknebel
I could reuse a bunch of stuff from the private webmentions for that kind of logic
#
aaronpk
yeah that's my thinking too
#
sknebel
just needed a different way of getting tokens
#
sknebel
started tracking questions that come up on the nürnberg page too. feel like some stuff was talked about, but not with everyone and not documented
[calumryan] and snarfed joined the channel
#
sknebel
... and just thought of another one...
[pfefferle], [Vincent] and [tantek] joined the channel
#
swentel
ok going to start testing the new built-in webmention endpoint in the drupal module
#
swentel
*fingers crossed*
#
swentel
aaronpk, added some classes which extend on XRay for now to handle u-follow ;)
#
aaronpk
nice yeah I should do that
#
swentel
aaronpk, I've added a pull request that should do it
#
aaronpk
cool I will add that to my list :)
#
swentel
might need to add tests, haven't checked
#
swentel
there's one open for aperture too, but I don't really need it anymore hehe :)
#
swentel
or was it watchtower, can't remember
#
swentel
ah watchtower yes
#
aaronpk
I have too many projects lol
#
aaronpk
as soon as I finish a round of updates/bugfixes to one, a different project has become overdue for attention
#
swentel
hehe, I know the feeling
#
swentel
I've also disabled all e-mail notifications for repo's everywhere, less stress :)
#
aaronpk
if I had a good way to look at issues across all my repos I would do that too
#
aaronpk
but github doesn't really provide that view
#
swentel
hmm I do use /issues for that
#
swentel
but I have less repos, so that's why it still works for me I guess
#
aaronpk
is there a magic search term that will show all open issues across all my repos? the default view includes author:aaronpk which is not what I want
[Rose] joined the channel
#
[Rose]
I think the app CodeHub on iOS can do that
[dave] joined the channel
#
aaronpk
I want to keep up on some repos in the indieweb org too
#
[Rose]
Ah, yes. I was wondering what the next step is for the checkin format. Only I should ask Tantek, it's his repo 😛
#
Loqi
[RosemaryOrchard] #40 Check ins as a specific property
#
aaronpk
I already use the "watching" feature to decide which ones to get emails about, why can't that setting also apply to a view on the web? that's what I really want
#
sknebel
there's also this https://octobox.io/, but it's based on notifications you got, not sure if it works quite right
#
aaronpk
oh yeah and https://github.com/notifications is also close but again not a very actionable view
#
aaronpk
I want a view like https://github.com/issues but that only includes repos in https://github.com/watching
[pfefferle] joined the channel
#
swentel
hmm advanced search seems to handle that
#
swentel
but the view is annoying
#
sknebel
is digging through UMA again
snarfed, sebsel, eli_oat, swentel and [eddie] joined the channel
#
leg
1033040575_pmDk2-
#
snarfed
i've always thought so too
#
jalcine
how do y'all handle thumbnailing on sites? I'm going to be making extensive use of them for performance concerns but I don't _think_ it should be a big idea mf2 parsing wise
#
jalcine
it'd be just a lot of the same image in different resolutions
eli_oat joined the channel
#
KartikPrabhu
jalcine: I use srcset on <img> but mf2 parsers will pick up only the src attribute
#
jalcine
ah okay
#
sknebel
yeah, srcset parsing has been proposed but since there has been no feedback at all on what output people would like I haven't followed it further
#
aaronpk
authorship question... I continue to miss out on any authorship information in adactio's posts from his feeds. In his feed pages, there is no author property on the individual h-entrys. His pages are a series of h-entrys with one h-card at the end. Should I add a rule to use that h-card as the author of every h-entry? The rule would be basically if the page consists of a bunch of h-entrys followed by
#
aaronpk
one h-card, and the h-entrys don't have a URL, use the h-card as the author of each.
#
swentel
that makes sense to me
#
sknebel
he's got a rel author too
#
sknebel
although it doesn't point to the home page
#
aaronpk
hm yeah that is tricky, especially since the url of the h-card on https://adactio.com/about/myself is https://adactio.com/
#
Loqi
Jeremy Keith
#
aaronpk
I suppose rel=author for a feed page could make some sense to use
#
aaronpk
really need a new page to document the feed authorship algorithm
#
gRegorLove
Is it a representative h-card?
#
Loqi
gRegorLove: [grantcodes] left you a message 8 hours, 18 minutes ago: I've not used IBC extensively yet myself, but the way I would imagine updates working would be practically the same as a new post, but if the user selects they have finished or got to page you could show an extra box to paste a post url to update.
#
Loqi
gRegorLove: [grantcodes] left you a message 8 hours, 17 minutes ago: also seems I can't login anymore. The auth redirect has a double slash in it which I'm sure doesn't help
#
aaronpk
what is representative h-card?
#
Loqi
The representative h-card for a page is an h-card on that page that represents that page, if any, as not all pages are about a person or organization, a page might not have a representative h-card https://indieweb.org/representative_h-card
#
aaronpk
sounds like no
#
aaronpk
that is a terribly worded sentence tho lol
#
sknebel
that's another thing that could take another look... since indiewebify.me promotes it so much people spend time on it, and I'm not even sure the definition is clear or it's actually consumed
#
aaronpk
it is definitely consumed
#
sebsel
what is authorship?
#
aaronpk
but the authorship algorithm is probably used more, and used before the representative h-card would be used
#
Loqi
authorship is how to determine who the author(s) of a post are https://indieweb.org/authorship
#
gRegorLove
indiewebify.me has a bug validating rep. h-card btw
#
sknebel
aaronpk: great, then please add examples to the section I just added :D
#
gRegorLove
aaronpk, what's the adactio feed URL in question?
#
Loqi
Jeremy Keith
#
gRegorLove
(disregard php notices, heh)
#
gRegorLove
The h-card there has a u-url that matches a rel-me on the same page, so representative
#
sknebel
https://adactio.com/about/myself/, which is the rel=me also validates
#
Loqi
Jeremy Keith
#
gRegorLove
the bug actually isn't in indiewebify.me but Barnaby's mf2 cleaner. I have a PR: https://github.com/barnabywalters/php-mf-cleaner/pull/6
#
Loqi
[gRegorLove] #6 Update the getRepresentativeHCard() method
#
sknebel
so if authorship would do an analog to step 6 for feed pages, it'd find it
#
aaronpk
so which is the better option for adactio's site? use the rel=author value, fetch that, and use the representative h-card on that? or look for the h-entry + h-card pattern?
#
sknebel
(well, if it accepts a page full of entries as a feed)
#
aaronpk
ah yeah it is kind of analogous to step 6
#
gRegorLove
er, that PR may not be related after all. I'll have to look at the code closer.
#
gRegorLove
It sounds like rep. h-card algorithm is easier in this instance, no extra fetch
#
aaronpk
no that is mixing the two options
#
aaronpk
option 1: fetch rel=author and use representative h-card to decide whether to use that page as the author on all the h-entrys
#
aaronpk
option 2: no additional fetch needed, just use the single h-card that's at the bottom of the page
#
sknebel
option 3: option two, but only if it is a representative h-card
#
aaronpk
I would feel better about option 2 if the rel=author value matched the url of that trailing h-card
#
gRegorLove
Yeah, option 3 is what I'm proposing
#
aaronpk
I thought representative h-card applies only to a page that is about a person
#
aaronpk
which means applying it to a feed page doesn't make sense
#
aaronpk
that was my understanding given that (poor) definition on the page
#
sebsel
or organisation*
#
sknebel
mine too, but in practice people put h-cards meeting those criteria all over the shop
#
aaronpk
"an h-card on that page that represents that page"
#
aaronpk
but an h-card on a feed page doesn't *represent* the page, it is the author of the entries on the page
#
sknebel
yeah, true, so it's not a good basis
#
KartikPrabhu
right. I don't think representative h-card is supposed to decide the "author" for h-feeds
#
sebsel
yeah, on a feed the h-feed would *represent* the page.
#
sknebel
true, at least the rel=me thing case shouldn't happen on random pages
#
gRegorLove
I guess I'm proposing option 4: follow /authorship algorithm; if not found, follow rep. h-card.
#
sknebel
homepages are a bit of a special case which often are both a feed and representing a person
#
gRegorLove
Fitting rel-author in there somewhere is probably worth exploring
#
aaronpk
gRegorLove: I just don't think it's a good idea to say you should use the representative h-card algorithm on something that is not intended to represent a person or org
#
aaronpk
probably a few of the steps in the representative h-card algorithm would be good to include in the feed authorship algorithm, but they are different things
#
gRegorLove
hm, maybe
#
gRegorLove
I think /authorship was originally based on rep. h-card
#
gRegorLove
Doesn't authorship algorithm 7.3 work with https://adactio.com/about/myself/? There's one h-card there with url = rel-me
#
Loqi
Jeremy Keith
#
gRegorLove
maybe not based on 6 "if there is no author-page and the h-entry's page is a permalink page" since this is a feed page. hm
#
gRegorLove
could add a step 4.1 "otherwise if the feed page has a rel-author, use that"
#
gRegorLove
I think that would definitely take you through step 7 then and find the h-card
#
aaronpk
well the real problem is currently there is nothing documented for feed authorship
#
aaronpk
so we don't really have anything to add text to right now
#
aaronpk
i've just been winging it for Aperture right now, loosely based off of permalink authorship
#
gRegorLove
Sure, but we have at least some mention of feeds in /authorship
#
aaronpk
oh you're right, the authorship algorithm is the right place for this, since I *am* talking about finding the author of a post. it's just that the post isn't on a post permalink page, it's within a feed.
#
sknebel
I mean, it makes sense to have a feed authorship algorithm, and reference that. "if the post is in a feed for which an author can be discovered, assume this"
#
sknebel
make a new section or a new page
#
aaronpk
just reread that in the context of adactio.com and I get to step 8 and do not find an author
#
gRegorLove
Try it with my proposed 4.1
#
aaronpk
that could work, but then we need a way to know if the page is a feed page
#
gRegorLove
"has a parent h-feed"
#
aaronpk
which has been another challenge
#
aaronpk
but adactio.com does not
#
sebsel
so there would be something like: are there multiple h-entry:s at the top level? -> treat it as implicit h-feed; are there multiple h-entry:s and one h-card -> treat it as implicit h-feed with the h-card as author of the feed
#
sebsel
and then in the discovery, at some point check if the feed has an author?
#
aaronpk
something like that, which is kind of why I was thinking this should be written differently than the permalink authorship discovery
#
sebsel
could still be on the same page tho.
#
gRegorLove
Think we'll need some url/uid/rel-me checks with "multiple h-entry:s and one h-card -> treat it as implicit h-feed with the h-card as author" e.g. a stream of h-entrys and only h-card is a link in one of the h-entrys to someone else.
[schmarty] joined the channel
#
gRegorLove
o/ [schmarty]
snarfed, Samnz and eli_oat joined the channel
#
Loqi
[tantek] #7 Should u-* parsing special case img srcset?
#
sknebel
snarfed: exactly. wish for an output format and you'll have a PR within a week, but I don't want to invent an output format I don't consume
#
sknebel
(my proposed "just a string" doesn't feel right, but I'm not sure what easy to consume as a more complex, parsed out format)
#
sknebel
(dict of "size rule":"url"?)
#
sknebel
(list of pairs or ordereddict, sorted by size?)
mattl, willnorris and voxpelli joined the channel
#
sebsel
Hm, working on Leesmap, hooking up Monocle as frontend, but I can't view certain channels because they are not in the channel list. :P
#
sebsel
It makes sense. Should we be explicit that all channels need to be in the channel-list?
#
sebsel
Meh, this is a weird edge case anyway, where I happen to know the identifier of a channel that is not in the list
#
swentel
and why is it not in the list ?
#
sebsel
because I have not added them.
#
sebsel
but also, because it's a channel that is created as an extra channel, on the fly
#
sebsel
(specificly: the 'liked' channel, which contains all the posts that are liked by posts in your feed)
#
sebsel
I should just add those to the list.
#
swentel
oh like that
#
swentel
interesting case
#
Loqi
agreed.
#
sebsel
no the longer I think of it, the more I think I'm just wrong. It would be nice if monocle would accept more channel identifiers, but it's totally fine it doesn't.
#
swentel
yeah, depending on the id, you decide on the server which posts it shows
#
swentel
it kind of that that on the fly too with notifications
#
swentel
hmm, s/that that/do that
[manton], snarfed, [jgmac1106] and [dave] joined the channel
#
sebsel
Oh, this is so satisfying... booting up one app (in this case Monocle) and finding out it more or less works with the code you just wrote (in this case Leesmap)
#
jalcine
regarding srcset and the like
#
sebsel
only had an issue with aaronpk's use of the short <?-syntax in php, but that was only related to Monocle, not to Leesmap's Microsub :P
#
jalcine
it doesn't change the case with mf2 parsing
#
jalcine
but it looks like that'd be a bit simpler to handle parsing (b/c you gotta do some sizes parsing with srcset and the sizes keyword)
#
KartikPrabhu
jalcine: the picture element also supports srcset and sizes in the <source>
#
KartikPrabhu
the <picture> by it self is just a wrapper element essentially
#
jalcine
right but I could have it work with the <img> as a "base" and not crowd the img tag
#
jalcine
(I'm thinking about HTML generation as well to be frank)
#
KartikPrabhu
sure by mf2 parsers don't really care if the <img> element is "crowded"
#
KartikPrabhu
so yeah you should go for what makes sense as the markup for your site and not worry too much about mf2
#
KartikPrabhu
in fact I use <picture> with <img> with srcset and sometimes <source> if I am using svg
#
KartikPrabhu
if you put say "u-photo" on an <img> then mf parsers will directly pick up the src attribute which work fine for now
snarfed and [dave] joined the channel