#indieweb 2016-09-04

2016-09-04 UTC
#
KartikPrabhu
bear: thanks. good to know people are looking at this
thebaer joined the channel
#
aaronpk
heh i usually use pin13.net or intel.com for that
userXYZ, Gold, tantek, singpolyma, KevinMarks_, KevinMarks, awolf and wolftune joined the channel
#
tantek
KartikPrabhu: I use example.com for triggering those captive portals.
pl_, begriffs, KevinMarks_, mlncn, KevinMarks, tantek, wolftune and chimo_ joined the channel
#
aaronpk
example.com is a good idea. /me files away a mental note
#
GWG
I've been reading the mention-client.
#
GWG
Trying to address some questions I have about sending webmentions
#
aaronpk
has it been helpful? I believe it passes all the relevant webmention.rocks tests
#
GWG
aaronpk: It's a lot of nitpicky things I'm trying to figure out
#
GWG
For example:
#
GWG
WordPress matches all URLs in text for Pingback.
#
GWG
The current Webmentions plugin only matches URLs in links(a href...)
#
GWG
And the mention client does it three different ways
snarfed, mblaney, KevinMarks, tantek and [kevinmarks] joined the channel
#
[kevinmarks]
Should it mention urls in author hcards of comments? That seems like a no.
#
GWG
[kevinmarks]: It only does it in the post_content block. So that isn't an issue
#
mblaney
haha for anyone managing their own servers, this is great: http://xkcd.com/1728/
#
Loqi
rofl
#
mblaney
I thought it would be up your alley Loqi!
tantek, KevinMarks, wolftune, KevinMarks_, begriffs, snarfed, friedcell, gkbrk, cmal, rrix, Kopfstein, catsup and cweiske joined the channel
#
cweiske
I was a bit disappointed that there are no working micropub cli clients
dmaczka joined the channel
#
cweiske
!tell aaronpk re micropub draft: why is the media endpoint only discoverable after fetching an auth token?
#
Loqi
Ok, I'll tell them that when I see them next
#
cweiske
does someone know how indieauth is supposed to work with CLI clients that cannot specify a callback URI?
gkbrk joined the channel
#
voxpelli
cweiske: like all other OAuth-based CLI clients?
#
cweiske
I never used a oauth cli client
#
voxpelli
cweiske: one way is to manually specify a token somewhere
#
cweiske
where do I get that token from?
#
voxpelli
cweiske: depends on what generates the tokens for you – if it's not your own system but rather eg. IndieAuth, then you would need a web flow that grabs it for you
#
cweiske
currently indieauth requires a callback URI to which the browser is redirected
#
cweiske
cli clients don't have such a redirect URI
#
cweiske
that's my problem
#
cweiske
voxpelli, any indieauth server that follows the current spec
#
voxpelli
cweiske: on eg. GitHub one can get a token here: https://github.com/settings/tokens
#
cweiske
voxpelli, then people could only use my client if they have an endpoint which supports such things
#
cweiske
manual tokens
#
voxpelli
cweiske: that just describes a way to discover and negotiate the way to a token – one can just as well get a token in another way
#
cweiske
I'd like some process that works on all indieauth servers
#
voxpelli
yeah, or you could make a web service that just echoes back the discovered token
#
cweiske
I'm doing that right now
#
cweiske
since I'm hosting that, I'm the one who gets access to everything
#
cweiske
which is .. not optimal, security-wise
#
cweiske
maybe indieauth should specify some special redirect_uri value that tells the auth server to echo the auth code
#
cweiske
so auth servers implement that themselves, and I don't have to host a service for other people only because I wrote a MP client
#
voxpelli
there are alternative flows to negotiate OAuth tokens that are more suitable for eg. CLI tools, TV:s etc – such as a pin-based approach, where the callback url is replace with a pin code
#
cweiske
ok. unfortunately, the indieauth spec does not allow such things
#
voxpelli
simplest change to make it support it would probably be to just make the redirect_uri optional in the authorization flow and if not specified then just print the code instead
#
@LukasRosenstock
What's it like to attend an @indiewebcamp? Read this IWC NYC2 wrap-up by @schmarty: http://buff.ly/2bOQgWx #indieweb
(twitter.com/_/status/772419204083490817)
seekr and cmal joined the channel
#
cweiske
voxpelli, the "code" currently looks like this: code=ZW1vamk9JTVDMzYwJTVDMjM3JTVDMjIyJTVDMjUxJm1lPWh0dHAlM0ElMkYlMkZhbm93ZWNvLmJvZ28lMkZ1c2VyJTJGMy5odG0mc2NvcGU9JnNpZ25hdHVyZT1GSVhNRQ%3D%3D&me=http%3A%2F%2Fanoweco.bogo%2Fuser%2F3.htm&state=1472992410
#
cweiske
three parameters: code, me and state
#
voxpelli
cweiske: only the code query param is needed though (maybe move to #indieweb-dev?)
[kevinmarks] joined the channel
#
[kevinmarks]
This sounds like it needs to learn about mf2 https://getpop.org/en/
seekr joined the channel
#
GWG
Does anyone return json in a Webmention implementation?
seekr joined the channel
#
bear
I know I would welcome an IndieAuth way to use my command line tools - right now my CLI says "after the browser opens, copy the access token..." which is not very helpful
#
cweiske
bear, i'm doing that, too
#
cweiske
but I'm evaluating the option to open a http server from my cli tool on a port and then use 127.0.0.1:12345 as redirect uri
#
bear
that reminds me of how the google cloud tools do things
cmal, wolftune, dmaczka, mlncn, miklb and tantek joined the channel
#
aaronpk
this is what the OAuth device flow is for
#
Loqi
aaronpk: cweiske left you a message 3 hours, 37 minutes ago: re micropub draft: why is the media endpoint only discoverable after fetching an auth token?
cmal joined the channel
#
voxpelli
aaronpk: I wonder if one could discover that a service supports that? and somehow use that?
wolftune joined the channel
#
GWG
I'm thinking of completely taking over the WordPress system for sending pingbacks and switching to a webmention first, pingback fallback design.
snarfed joined the channel
#
aaronpk
can you do that? that sounds like a good idae
#
GWG
aaronpk: Yes.
#
GWG
The current plugin sends webmentions independently of pingbacks.
#
GWG
Then adds the mentioned URLs to the don't pingback list.
#
GWG
But the design I'm looking at is what it would look like if webmentions were supported in Core.
#
snarfed
aaronpk: http://www.webmention.io/ is just showing the nginx start page right now. intentional?
#
GWG
By the way, aaronpk...the ability for a WordPress site to delegate its pingbacks to webmention.io should be in WordPress 4.7. Until now, you couldn't filter the default pingback endpoint easily. Is that a problem?
#
snarfed
GWG: bridgy respond json to wms, both for hosted blogs and for publish
#
GWG
snarfed: As long as I'm using the REST API, which autoconverts every output to JSON unless you turn it off, I figured I'd consider outputting as JSON.
#
GWG
It doesn't seem to be forbidden in the spec
#
aaronpk
snarfed: no www :(
#
aaronpk
i'll add a redirect
#
snarfed
aaronpk: ah thx
#
snarfed
GWG: webmention.herokuapp and wm.io (i think?) might too
#
GWG
Saves me a bunch of code.
#
voxpelli
GWG: snarfed: I send JSON on some errors, but on most responses only send JSON or plain text if stated as an acceptable response format by client
tantek joined the channel
#
GWG
voxpelli, what do you think makes sense?
#
voxpelli
GWG: json sounds good
wolftune, Loqi and awolf joined the channel
#
tantek
hmm - does this count as a silo quit? or perhaps just a silo reset? http://mashable.com/2016/04/08/saint-laurent-instagram
#
tantek
"Why Saint Laurent abruptly deleted its entire Instagram history"
#
bear
I would say silo reset
#
bear
they haven't *left* instagram, just did a "winner (re)writes the history" on it
#
tantek
I feel like I've seen other people do similar "wipe and start over" on silos
#
aaronpk
Happy birthday Loqi!
#
aaronpk
Loqi: a/s/l
#
Loqi
252465131/bot/internet
#
aaronpk
252465131 seconds = 8.005 years
wolftune, KevinMarks and [kevinmarks] joined the channel
#
[kevinmarks]
What do you mean by json response form webmention? The html form an endpoint is not usually shown, so what's the json.
#
[kevinmarks]
Or do you mean for output from webmention.herokuapp.com or webmention.io to construct wm responses for your site?
#
Loqi
[indieweb] "Comment on Representing Indieweb at DrupalCampLA by IndieWebCamp" by IndieWebCamp on 2016-09-04 http://boffosocko.com/2016/08/27/representing-indieweb-at-drupalcampla/#comment-31991
#
GWG
[kevinmarks]: I am wondering if I can return error and success responses from a webmention endpoint in JSON as opposed to plaintext
#
aaronpk
absolutely
#
GWG
Yes.
#
GWG
The REST API infrastructure in WordPress is the 'new hotness' in endpoint construction for WordPress.
#
GWG
I have never followed the bandwagon before, but I have less to build myself.
#
KevinMarks
Yes, that should be fine
goodoo, tvn, pl_ and KevinMarks joined the channel
#
aaronpk
whoa "Alena added a temporary profile video."
myfreeweb joined the channel
#
Zegnat
Not going to lie, those animated profile pictures are pretty fun. Though to be fair, acegiak has been rocking one of those for some time now (hashtag-indieweb)
AndChat|566325 and goodoo joined the channel
#
@rachelandrew
Lanyrd disappearing all weekend has prompted me to work on adding speaking stuff to my own site. Amazing how much data I didn’t ‘own’.
(twitter.com/_/status/772432792529539072)
#
tantek
what is Lanyrd?
#
Loqi
Lanyrd is an event silo focused on conferences https://indieweb.org/Lanyrd
#
tantek
http://lanyrd.com "Ooops! / 503 / Gosh! Looks like something has gone a bit wrong. Don't worry though, we know and are fixing it. Please try again later. /Follow @lanyrd on Twitter for updates or email support@lanyrd.com"
wolftune, tantek and awolf joined the channel
#
tantek
!tell benwerd ICYMI re: POSSE: https://twitter.com/colinjohnston/status/772533151960707072 - when a silo is down, so is any PESOS approach. with POSSE, silos going down can't stop your site.
#
Loqi
Ok, I'll tell them that when I see them next
KevinMarks and pl_ joined the channel
#
GWG
I'm obsessing and need a perspective again
#
GWG
If you are looking for webmention and if isn't there, you should go for pingback, that means I would have to look in the head for webmentions, then the body, then the head again for pingback.
#
tantek
GWG, not necessarily
#
tantek
I have written code to do this without having to check the head twice
#
GWG
tantek: I intend to hold the head without retrieving it a second time. I'm wondering what the most efficient way to do it is.
wolftune, AndChat|566325, goodoo and snarfed joined the channel
#
snarfed
!tell tantek benwerd's email to known-dev wasn't about posse vs pesos. it was about posse + backfeed vs pure indie interactions.
#
Loqi
Ok, I'll tell them that when I see them next
mlncn, tantek and begriffs joined the channel
#
Loqi
tantek: snarfed left you a message 15 minutes ago: benwerd's email to known-dev wasn't about posse vs pesos. it was about posse + backfeed vs pure indie interactions.
#
tantek
snarfed, then I don't understand, as they are complementary, not vs
#
tantek
GWG, did you find the code?
#
GWG
tantek: It isn't about the code. Trying to decide how to put it together.
#
tantek
hmm I suppose an outline of how to call it would help
#
GWG
Every time I map it out, I feel like I'm repeating myself.
#
tantek
that library helps reduce repeating yourself, that's the point
#
miklb
can't remember if ASL is compatible with GPL
#
GWG
tantek: Not really, alas. The code for extracting the information is already in WordPress. I'm just trying to figure out how to gracefully fall back.
snarfed joined the channel
#
snarfed
tantek: try reading the email again?
#
snarfed
(sorry, not trying to be snarky :P)
#
tantek
snarfed, the whole premise of "APIs changing breaks POSSE" seems like arguing an absolute in a vacuum
#
GWG
I just want to make sure it doesn't get criticism.
#
miklb
as long as its constructive, it would only make it better
#
tantek
snarfed, also really curious about how you got the "letter of the law" impression when /principles and /code-of-conduct would hopefully give a different impression?
#
KartikPrabhu
GWG: ask forgiveness not permission. If anyone criticises it then think about changing
#
snarfed
tantek: the api fragility point was a smaller point, just at the beginning. the bigger point was surprise and social norms around backfeed
#
snarfed
for people who *aren't* part of the community, brought in via backfeed
#
KartikPrabhu
tantek: the one thing I got from that email is that silo users might not realise that their responses are backfed. That seemed to be the most relevant point
#
tantek
points about surprise and social norms are certainly worthy of more documentation
#
snarfed
that was the primary point
#
tantek
especially when there are specific instances to be documented
#
KartikPrabhu
hence I put a warning in my Twitter bio
#
tantek
KartikPrabhu: did you add it to /disclosure#IndieWeb_Examples ?
#
tantek
KartikPrabhu: as did I with my FB profile
#
snarfed
others do too! not sure that's nearly enough though
#
KartikPrabhu
what is disclosure?
#
Loqi
A disclosure is a bit of content, typically on a home page, on an indie web site that proactively discloses some aspect about the site that the site owner wants the user to explicitly be aware of https://indieweb.org/disclosure
#
tantek
snarfed, why is "nearly enough" the metric?
#
tantek
isn't any incremental improvement a good thing?
#
KartikPrabhu
oh hey I aleady have added it!
#
tantek
In fact I'll even say that "not nearly enough" thinking is how you talk yourself out of shipping
#
tantek
as opposed to "good enough for now"
#
KartikPrabhu
snarfed: I agree that it might not be enough since the Twitter bio page is rarely visited (at least by me) when replying
#
snarfed
social norms are a very different thing than shipping code
#
snarfed
anyway, gotta run, definitely worth talking more later!
#
tantek
no successful social network started with thinking "not nearly enough"
#
tantek
so yeah, I'm going to reject that kind of framing, because it causes you to not ship
#
tantek
KartikPrabhu: it's never enough. that's the problem. you can always think of more fear-based excuses.
#
miklb
so if I'm understanding the conversation, people are bothered their public silo comments that are used to mine data to sell ads are bothered the comment might be aggregated with the original comment on a personal site?
#
tantek
thus "not nearly enough" is nearly useless framing
#
tantek
miklb - no, the assertion is, they *might* be
#
tantek
that's the problem
#
KartikPrabhu
tantek: yes, of course. I was simply wondering if there is a solution currently; and I didn't come up with anything
#
tantek
hence the "not nearly enough" assertion
#
tantek
KartikPrabhu: of course there's no solution
#
tantek
there's never a solution for "not nearly enough" thinking
#
tantek
the very thinking itself is flawed
#
tantek
as soon as you give into "not nearly enough" reasoning, you've already lost
#
KartikPrabhu
no. I was thinking of the problem: "Is there a way to warn users, while they reply on a silo, that their comment might be backfed"
#
KartikPrabhu
and the current answer to all silos I use is "no"
#
miklb
seems silly me. like thinking your conversation in a crowded bar is private.
#
tantek
KartikPrabhu: what is the utility of such warnings? what's the use-case? (especially given that usually such warnings just get ignored by users, like nearly all warning dialogs)
#
tantek
KartikPrabhu: the question "Is there a way to warn users?" presume that's a good UX, which it is not
#
tantek
miklb - that's a different problem, and somewhat more nuanced, see /publics for more on that
#
KartikPrabhu
the utility is to give fair warning (for sake of politeness) that their content may appear where they might not expect it to.
#
miklb
right, I realize it's not perfect analogy.
#
KartikPrabhu
I would want to know if my content appears somewhere I don't expect
#
KartikPrabhu
same as "I might use your content for advertising"
#
aaronpk
I don't know, frankly anything you put on the Internet in public you shouldn't be surprised about finding elsewhere
#
aaronpk
private or semi private is different for sure
#
KartikPrabhu
no. I am not surprised, but it is good to know before hand
#
tantek
KartikPrabhu: there is no utility when such warnings are known to be ignored in practice
#
tantek
you are assuming that giving users such warnings actually has any effect
#
tantek
which has been shown to be false by study of warning dialogs
#
KartikPrabhu
it does affect me. Maybe not most users.
#
KartikPrabhu
I am not going by plurality here
#
tantek
KartikPrabhu: I disagree. Since everything you post (what aaronpk said) would need such a warning, and currently you don't get such warnings for everything you post.
#
tantek
and you still post.
#
tantek
therefore it must not be affecting you.
#
tantek
this is also one of the reasons that when Bridgy was first proposed, there was a strong consensus to only backfeed replies to public posts
#
KartikPrabhu
yes. backfeeding only public posts is an acceptable solution in my mind.
#
tantek
snarfed, and I have discussed the implications of backfeeding responses to private posts, and have yet to come up with a way that would make the privacy expectations make any sense (be predictable)
#
tantek
so no one is doing it yet AFAIK (backfeeding responses to private posts)
#
aaronpk
Would sharing the same privacy settings (visibility?) be appropriate?
#
tantek
if there was some way you could pull that off!
#
tantek
cross-site ACLs!
#
aaronpk
yeah that's a can of worms
#
KartikPrabhu
what is ACL?
#
Loqi
ACL is an abbreviation for access control list, and in the indieweb context refers typically to a list of people who are allowed to view (and perhaps edit etc.) particular posts or perhaps an entire site https://indieweb.org/ACL
#
aaronpk
say that gives me an idea
#
tantek
yes if you had some way of rel=me determining that the same person would have access, then you could do indieauth based ACLs
#
tantek
tough to dynamically keep consistent though
#
aaronpk
That wasn't my idea ;-)
#
aaronpk
the classic problem is that sites with private posts have two different ways of handling access control. In one case, people who are added can see things previously posted to that group. In the other case, people who are added can only see things posted from that point onwards.
#
KartikPrabhu
is there any real need for consistency of the above across sites?
#
KartikPrabhu
same way we do not require consistency in say webmention handling or parsing?
#
tantek
aaronpk: I think nearly everyone has settled on the first model
#
tantek
once you get access, you can see all history
#
tantek
e.g. private twitter, IG, etc. accounts
#
tantek
once your request to follow is approved, you see everything back in time
#
tantek
same it true with e.g. W3C member-only mailing lists
#
tantek
so the other model is fairly ignorable (unless you have specific examples you want to document)
#
tantek
what is private?
#
Loqi
private posts refer to posts or portions of posts which are private to either the author or to a limited audience chosen by the author https://indieweb.org/private
#
tantek
what is a private account?
#
Loqi
A private account is a silo account where all posts on its profile are private, and only approved followers (or friends) can see posts https://indieweb.org/private-account
#
aaronpk
private IRC channels (or other things that are not logged) are the other way. But that does seem less common
#
tantek
aaronpk - that's only because of the accident of IRC logs being separate from chatting in them
#
tantek
presumably once you get access to the channel, you can see all logs too
#
tantek
w3c operates that way
#
tantek
slack is the same way. once you get access to a slack, you can scroll back to the beginning of a channel
#
tantek
anyway, for practical purposes, privacy models appear to have converged
#
tantek
for anything with a permalink anyway
#
tantek
I do think this (private posts) is an area indieweb is going to have to innovate soon to continue broadening reach / appeal (assuming that's a goal)
#
tantek
I'm certainly hesitant to implement some post types without it (e.g. checkins)
pl_ joined the channel
#
rrix
aaronpk: tantek: matrix.org supports both models with room history, on a per-room basis
#
rrix
ah heck, that URL probably mangled
#
aaronpk
rrix: wow that's thorough