#indiewebcamp 2015-07-15

2015-07-15 UTC
#
GWG
I think I fixed that in May, but does anyone speak French?
#
kylewm
GWG: it doesn't say anything specific, just that input isn't filtered before being inserted into the output HTML
#
GWG
That's not really helpful.
#
aaronpk
huh, i was writing out my thoughts on how to do private webmentions, but then realized I had already written this all out
#
GWG
I rewrote that code after the date in question.
#
GWG
Either way, better recheck it
KartikPrabhu joined the channel
#
GWG
When I decided to get into Indieweb development, no one said I'd have trouble with French security researchers
TysonNW joined the channel
#
KevinMarks_
no-one expects the French Inquisition
#
aaronpk
okay. who here is interested in brainstorming on private webmentions with me
snarfed1 joined the channel
#
GWG
aaronpk: I'm not sure what I have to contribute, but if something occurs, I'll chime in
#
KevinMarks_
sure, fire away aaronpk_
#
aaronpk
i'm mostly interested in which of two options you would be more likely to implement
#
bret
im listening but I'll need a dynamic messaging server first
#
aaronpk
jeez this is from almost exactly two years ago
#
aaronpk
okay here's the situation. say you wrote a private post that I can see. I'm making a GET request to fetch the post.
#
aaronpk
the options are:
#
aaronpk
a) I send a header like 'Authorization: IndieAuth me="https://aaronparecki.com" token="1234567890"', you look at aaronparecki.com and find my token endpoint, then check my token is valid by making a GET request to my token endpoint
#
aaronpk
b) before-hand, I make a POST request to your token endpoint with an authorization code that I generated, you check the code against *my authorization endpoint* and if it's valid, you generate an access token. then to actually fetch the post, I make a GET request with the access token that *you* generated in the Authorization header
#
aaronpk
does that make sense?
#
aaronpk
the key difference is that in b, you generated the access token so you can verify it however you want
#
aaronpk
with a, I generated the access token so you ahve to ask my token endpoint whether it's valid
#
bret
i dont know which is better
#
bret
how does pump.io do it?
#
aaronpk
doesn't matter which is "better"
#
aaronpk
what matters is which would you be more likely to implement
#
aaronpk
my question is which of a or b would you be more likely to implement, based on ease of implementation or whatever
#
bret
how are you making the get request?
#
bret
from a browser?
#
aaronpk
doesn't matter
#
aaronpk
but more likely this is server-to-server interaction
#
GWG
Should I have to satisfy you...the person who wants it private?
#
KevinMarks_
so the assumption is that I have an ACL that has 'https://aaronparecki.com' in
#
aaronpk
GWG: you wanted it private
#
aaronpk
KevinMarks_: yes
#
GWG
Then you should have to satisfy me.
#
aaronpk
i agree :)
#
aaronpk
which is why i am asking
#
KevinMarks_
so the implementation difference is whether I have to do work to create a token for you and keep it with that
#
aaronpk
KevinMarks_: i guess that's one way of putting it
#
KevinMarks_
or if I trust your token service
#
bret
do self encoded tokens change that at all?
#
aaronpk
bret: self-encoded tokens are certainly one way of implementing tokens, but should not be a requirement
#
KevinMarks_
I think a) is easier, as I don't have to store access tokens for you
#
snarfed1
+1 for a)
#
KevinMarks_
though I do have to do the check synchronously?
#
aaronpk
ooh yes you would, because you'd need to return the content in that response
KartikPrabhu joined the channel
#
aaronpk
KevinMarks_: interesting point about trust, if you're writing something that you want only me to see, you have to trust that my token endpoint will respect that. is that a reasonable expectation?
#
KartikPrabhu
aaronpk, snarfed: I saw some chat about hfeed2atom as a proxy service. Right now it is only a lib (and an incomplete one ). Is a proxy service something of interest?
#
aaronpk
(KartikPrabhu: I won't have an atom feed until there is a proxy service that I like, so yes. but i'm gonna keep focusing on this webmention conversation for now)
#
aaronpk
okay so i'm hearing some amount of preference for not wanting to have to generate/store access tokens
TysonNW joined the channel
#
KartikPrabhu
aaronpk: ok cool
#
snarfed1
aaronpk: KevinMarks: the auth token would be for the auth server specified in the page's rel=token_endpoint, right? so not necessarily indieauth?
#
snarfed1
so then servers that are latency sensitive can be their own auth/token server, or use one nearby
#
snarfed1
goddamn 1
#
Loqi
hehe
#
KevinMarks_
no it's the viewer's token endpoint snarfed
#
aaronpk
snarfed: indieauth.com is not in this picture unless someone explicitly links to it
#
KevinMarks_
not the publisher's
#
snarfed
ohh right since the viewer is asserting who they are
#
aaronpk
but yeah it's the other person's auth server
#
aaronpk
so you can't guarantee the responsiveness of it, or even if it still exists
#
KevinMarks_
so the advantages of b) is that the publisher just has to check it's own token, which it can have done async
#
aaronpk
with b, the publisher generated the token so it has an instant way to verify it. either a DB lookup or verifying the self-encoded token.
#
ben_thatmustbeme
tries to catch up on the conversation
#
KevinMarks_
so that's the tradeoff - do I want to store things for you, or do I want to make 2 roundtrips to your servers to respond to your GET
#
KevinMarks_
you have to have a long enough timeout to cope with that
#
aaronpk
i think there's just one round-trip? but yes
#
KevinMarks_
once to get your endpoint, once to verify the toekn
#
aaronpk
hm, requiring a sync HTTP request in order to respond to an HTTP request seems sketchy
#
aaronpk
KevinMarks_: oh right thanks
#
ben_thatmustbeme
wonders if his work on micropub-chaining make link in here (link /chain pun, heh)
#
aaronpk
okay i am leaning towards b now
#
GWG
ben_thatmustbeme: Is your Micropub code up?
#
aaronpk
also KevinMarks_ there is a way to issue access tokens where you don't need to store anything
#
ben_thatmustbeme
GWG, not working, i got an error on last try and haven't had time to fix the bug
#
aaronpk
also these tokens only need to be valid for like 30 seconds tops so that makes it particularly easy
#
GWG
ben_thatmustbeme: Up as in viewable by me.
#
ben_thatmustbeme
github.com/dissolve/mptweet
#
ben_thatmustbeme
and github.com/dissolve/openblog
#
ben_thatmustbeme
obviously requires support on both sides
#
ben_thatmustbeme
really disorganized at the moment
#
ben_thatmustbeme
idea is that site A needs to get a token that service X
#
GWG
ben_thatmustbeme: I'm looking at the WordPress Micropub code and looking for inspiration
#
ben_thatmustbeme
user logs in to service X (and does some other auth to make sure they have access to syndication target, i.e. twitter oauth)
#
ben_thatmustbeme
then service X uses indieauth to get a token for site A (the reverse of what is needed) but with a scope to add syndication endpoints
#
aaronpk
ben_thatmustbeme: any thoughts on this private webmention thing?
#
ben_thatmustbeme
once is has a token it sends and auth code to site A's mp endpoint telling it to register this new chaining target
#
ben_thatmustbeme
and finally, and this is where it ties in to the current discussion aaronpk, the site then just has to validate that code on service X
#
aaronpk
ben_thatmustbeme: that sounds like option a that I outlined?
#
ben_thatmustbeme
so basically my model is, send a code, and then it has to be validated back with the sender's endpoint
#
ben_thatmustbeme
aaronpk: yes
#
ben_thatmustbeme
and sorry, brain not working so well, but i have to try to stay up to keep my sleep schedule from getting too messed up
#
ben_thatmustbeme
which is probably a bad time to work on the thatmustbe.us service
Leeky joined the channel
#
ben_thatmustbeme
which is another question... i'm hitting some interesting problems with parsing people's h-cards
#
ben_thatmustbeme
like for example the lack of a single u-url on many of them
#
aaronpk
isn't the url of a homepage h-card the url that you got the h-card from?
#
ben_thatmustbeme
unless someone doesn't have the full card on their homepage
#
ben_thatmustbeme
bug i'm going to have to do that
#
sparverius
general channel question: how do you handle building assets from static site generators?
#
Loqi
ben_thatmustbeme meant to say: but i'm going to have to do that
#
sparverius
right now im pointing an nginx directive at a "built" directory, and build assets are in there, with jeykll assets in a directory in git root
#
sparverius
im using a pretty standard vps provider but eventually id like to include jeykll version management as part of the build process
#
sparverius
the standard vps provider being important because im hesitant to fire up a salt/vagrant vm to run nginx
#
sparverius
im overthinking this, but it's because it's a thing that interests me
#
ben_thatmustbeme
working on things that interest you ++
#
sparverius
oh yeah, ive migrated to doing builds on the vps as part of a cron job.
#
sparverius
previously i did the builds on my local machine, which i might go back to, since it reduces complexity of the server, which i would like to not have to think about
#
kylewm
sparverius: you could trigger the build from a git post-commit hook instead of on a cron
#
kylewm
(or you could have TravisCI build it...which I have done before)
#
sparverius
ah, what i meant is right now at midnight, git pulls, runs the build
#
sparverius
it's a hacked out bash in the crontab
#
kylewm
git hook would let you re-build as soon as you push changes, why wouldn't that be preferable?
#
sparverius
oh, yeah, im sorry, i misread what you said
#
ben_thatmustbeme
hmm, what would be more useful? thatmustbe.us/ben should return h-cards of all bens, but should it be regex to match *ben* or ben with lastname too or ben in url
#
sparverius
yeah, i'll add githooks to the list of things to implement
#
aaronpk
ben_thatmustbeme: I think libreral search is better, since firstname+lastname is not always guaranteed
#
sparverius
right now im focusing on keeping the software up to date, to ensure my dev and remote machine are on relatively similar versions
#
ben_thatmustbeme
hmm, okay, now i just need to work out a storage system
#
ben_thatmustbeme
going to try going without a database
#
ben_thatmustbeme
I'm hopping the fence for this one
#
ben_thatmustbeme
well storage and fast lookup without a DB
KartikPrabhu joined the channel
#
kylewm
aaronpk: still trying to parse, your idea (b) above is "the auth servers do a dance so that MY auth server gives you a token that identifies you as you to me"?
#
kylewm
that is pretty clever
#
aaronpk
it's actually a lot like "logging in" to someone's site
snarfed joined the channel
#
kylewm
I think that's what pump.io does, sorta
#
kylewm
you log into the remote server using your email address at the original server
#
aaronpk
i couldn't figure out what pump.io does, it just says "use oauth 1 or JWS"
#
kylewm
oh, yeah i just mean, as a user
#
aaronpk
okay then yeah i guess so
#
kylewm
the pump.io spec is pretty sparse and/or useless
#
kylewm
(which is not to denigrate pump.io)
lukebrooker joined the channel
#
aaronpk
making useful specs is actually pretty hard
#
aaronpk
it's always a tough balance between making things too specific vs making things not specific enough
cmhobbs, scor and KevinMarks__ joined the channel
#
aaronpk
is digging deep into the private webmention page
KevinMarks joined the channel
#
@kid_OYO
RT @NZN: Summer Tech Fun #youngmakers + $OYO skills #Tesla #pi #javascript #indieweb with @oyoclass @sbucompsc @TeslaMakerLab http://t.co/o…
(twitter.com/_/status/621148411023237120)
#
@nodenow
Summer Tech Fun #youngmakers + $OYO skills #Tesla #pi #javascript #indieweb with oyoclass sbucompsc TeslaMakerLab https://twitter.com/NZN/status/621147514473959424/photo/1
(twitter.com/_/status/621149500212662274)
KevinMarks__ joined the channel
#
aaronparecki.com
edited /private-webmention () "(-3864) update with latest on private webmention thoughts"
(view diff)
skalpin, GWG, lukebrooker and tilgovi joined the channel
#
cweiske.de
edited /Category:IndieAuth (+23) "/* Services and Source Code */"
(view diff)
#
aaronpk
thx cweiske
wolftune, tilgovi, chreekat and GWG joined the channel
lukebrooker joined the channel
wolftune, snarfed, tantek, squeakytoy, Deledrius, KartikPrabhu, cweiske, LauraJ, KevinMarks, KevinMarks__, Jihaisse, eschnou, loic_m, sanduhrs, jciv and csarven joined the channel
petermolnar, glennjones, modem, interactivist, keroberos, Sebastien-L, mapkyca, friedcell, shiflett, j12t, KartikPrabhu, kegsay, edpw1, tvn, evalica, stream7, csarven, Phyks, martinBrown, fkooman, frzn and scor joined the channel
#
@NoizIvy
RT @NZN: Summer Tech Fun #youngmakers + $OYO skills #Tesla #pi #javascript #indieweb with @oyoclass @sbucompsc @TeslaMakerLab http://t.co/o…
(twitter.com/_/status/621275797731545088)
LanceyWork, LauraJ, hardfire, j12t, KartikPrabhu, parzzix, scor, shiflett_, zachdono-, dpg1, shalkydri, friedcell and tilgovi joined the channel
#
cygnoir.net
edited /2015/Edinburgh/Guest_List (+252) "/* Regrets */"
(view diff)
#
csarven
Proposal to simplify Webmention: The Webmention endpoint receives a POST with the source URL, where its response contains the target URL (implying that this is the reference which needs to be confirmed). The endpoint can decide to whatever it wants with the rest of the contents at the source URL.
#
cweiske
so the endpoint first has to parse the source URL to reply?
#
cygnoir.net
created /User:Cygnoir.net (+177) "Created page with "__NOTITLE__ = cygnoir = Halsted M. Bernard == Background == I've been writing on the Web since 1999, but always with someone else's tools. This year I'd like to change that.""
(view diff)
#
cweiske
what if there are multiple target URLs on the source URL?
#
rhiaro
They all get a webmention? Since they're all mentioned
#
cweiske
they all would have to be returned
#
rhiaro
You have to parse the source to confirm the target is really there anyway
#
cweiske
in csarven's proposal
#
csarven
cweiske Yes, first parse to confirm. Good question about multiple target URLs, is that a common practice? If the target's domain matches the Webmention's endpoint (or whatever it finds appropriate), I suppose the endpoint will take those in.
#
cweiske
csarven, waiting for parsing means that means that my tools have to wait for each endoint to parse my site
#
cweiske
imagine a blog post with 20 links
#
cweiske
I ping all 20 endpoints
#
cweiske
and have to wait for all of them to parse
#
cweiske
they cannot defer parsing to later
#
cweiske
because they are required to return the urls
#
voxpelli
to have all URL:s of a HTML page be counted as WebMentions would make internal WebMentions explode :P
#
csarven
I didn't say "all URLs"
#
voxpelli
well, webmentions only say that a source page has linked to a target page, how would it know what URL:s to pick if it wouldn't pick all?
#
voxpelli
that would couple the h-entry and webmentions standards – right now they are completely independent of each other
#
csarven
cweiske is on track. I think the interesting bit to consider is whether there are multiple "in reply to"s or what have you, and make a decision.
#
csarven
voxpelli Right. But, I didn't suggest h-entry or mf2 or necessarily even "u-in-reply-to"
#
rhiaro
Wms are not limited to replies
#
rhiaro
Just links
#
csarven
Sure, whatever property.
#
csarven
x has relation to y
#
voxpelli
csarven: might have misunderstood you then, how did you suggest that one would distinguish between "mentioned" URL:s and other URL:s on a page if one doesn't mention it in the ping?
#
voxpelli
just as an endpoint today has to do a lookup of the source URL, the source has to do a lookup of the target URL – so the source will also have to know about all target URL:s anyway
#
csarven
If "mention" is the most appropriate term/relation to make between the source and target, that's fine. The endpoint will look for u-mention (if it is interested in mf2). It still leaves it open as to what it can look for.
#
cweiske
voxpelli, but today endpoints can delay parsing
#
csarven
voxpelli I suppose the target's endpoint can poke source's endpoint?
#
voxpelli
cweiske: yeah, my endpoint certainly delays it :)
#
cweiske
mine, too
#
Loqi
cweiske meant to say: mine too
#
csarven
The value of displaying without verifying is I suppose you get some real-timeness out of it.. but still subject to being an "unverified claim".
#
voxpelli
csarven: currently my endpoint treats all target URL:s that are included in the source page as a mention – no matter if there's mf2-data or not - mf2-data merely enhances the relation with additional data
#
cweiske
csarven, accepting the webmention does NOT mean displaying it
#
rhiaro
With an explicit target the endpoint has the option of disregarding it at once if it's not their domain, without having to parse source
#
cweiske
it only means that my endpoint has accepted it, and will handle it as it wishes
#
csarven
cweiske Fair enough, .. what does it mean then? Is it used in any way?
#
cweiske
yes. put in the queue
#
voxpelli
my endpoint is also shared between many sites so I would have a hard time knowing what target URL:s to look for if the source site doesn't tell me its intentions
#
cweiske
to be processed later
#
csarven
cweiske That's fine. I think that holds with my proposal. Accept the POST, decide what you want.
#
rhiaro
And knowing what to look for presumably just makes parsing easier
#
cweiske
no. I have to parse to return the URL
#
cweiske
"where its response contains the target URL"
#
voxpelli
not sure why it would be simpler to not tell the endpoint about the target URL as one would anyway have to look up the endpoint by fetching the target URL?
#
voxpelli
each and every target URL can have a different endpoint
#
csarven
Perhaps I'm not understanding the value of submitting the target URL if it is not "really" put to use. Isn't it true that the source URL has to be checked regardless before it is a confirmed mention?
#
voxpelli
One verifies whether the claimed mention is actually present in the source – but to verify a claim you have to have a claim in the first palce
#
csarven
What overrules? The POST information or what's in the source response?
#
voxpelli
the POST-data is the claim, the source is the verification, if the verification fails then the data is discarded
#
voxpelli
so one doesn't really overrule the other
#
csarven
Without verification, there is nothing to discuss.
#
csarven
Unless the endpoint is somehow putting the claims into use e.g., displaying.
#
csarven
Just to emphasize, I think it is important that the source URL is a permalink of the note/post/h-entry or whatever.
#
rhiaro
^ that's required right?
#
rhiaro
There's no case where it wouldn't be
#
cweiske
webmention has no dependency on microformats
#
rhiaro
The source just had to contain a link to the target
#
rhiaro
Not necessarily marked up
#
cweiske
webmention has no dependency on html
#
csarven
cweiske Right, and that's all good I think.
#
csarven
A "webmention" can use mf2 or RDFa or xyz.
#
cweiske
no. a webmention doesn't use anything
#
csarven
The target only needs to look for property/ies that it wants to look for. It is simply receiving a list of claims (subject URLs) to go and check. based on whatever criteria it finds appropriate, it can display or use those "statements" as it likes.
#
csarven
cweiske I realize that it doesn't use anything in particular. It is however implicitly doing that.
#
cweiske
no. the tools that built upon it do
#
voxpelli
csarven: I would rather say: an endpoint A receives a claim of one page, B, mentioning/linking to another page, C, which has chosen to use A to gather WebMentions
#
csarven
cweiske I don' tknow why you are in the negative.. perhaps I'm not clear. Yes, the apps/service/whatever you want to call it is looking for whatever it wants to look for.
#
voxpelli
"A" verifies that the claim that B links to C is true and then does whatever further processing it feels it wants
#
csarven
voxpelli Yes, that's it.
j12t, ttepasse, Jihaisse, fourtonfish, shiflett and snarfed joined the channel
#
Loqi
slack/snarfed: csarven: i think you might be right. i keep trying to come up with reasons target is required, but i can't
#
Loqi
slack/snarfed: at most, it changes the semantics a bit, from "process this wm for one target" to "process this wm for all targets you handle"
#
cweiske
there is a reason
#
cweiske
what if I have a page that contains 3 links to your server, but only one in the part of the page that I deem to be the content of the article
#
cweiske
my tool knows this and sends only one webmention for the one link
#
cweiske
but with csarven's proposal, the endpoint would do the parsing and find all three links
#
cweiske
because it does not know that I don't want to have the two others send to it
snarfed1 joined the channel
#
csarven
Which is why I emphasized on the point that the permalink (an item of some sort) is used as the source URL and not just any "page".
#
cweiske
that does not matter
#
cweiske
the source page contains 3 links
#
cweiske
and endpoint needs to find out which of the links on the page it wants to process
#
csarven
And only one of those links is of concern to the endpoint (as provided in source URL)
#
cweiske
source URL links to example.com/foo, and in its footer to example.com/a and example.com/b
#
cweiske
example.com's endpoint now needs to find out that *every* page in my source URL blog links to example.com/a and example.com/b
#
ben_thatmustbeme
cweiske: so what if it does, HOW to parse the data at that url is outside of this spec
#
cweiske
and that I don't want to send out webmentions for it
#
cweiske
only for the example.com/foo link
#
snarfed1
cweiske is right, this removes the ability of the source to selectively send wms for only some links
#
cweiske
but I can't tell this to the endpoint if there is no target url parametert
#
snarfed
but *anyone* can send a wm for any source and target urls, so that's not a strong feature anyway
#
ben_thatmustbeme
feels this should probably move back to #social since it is really departing any indiewebcamp discussion
#
snarfed
(ben_thatmustbeme: really? does more webmention discussion happen there than here?)
#
ben_thatmustbeme
snarfed: just thinking that if its moving toward a discussion without mf2 and how things work without mf2, not sure its really too important here
#
ben_thatmustbeme
there has been discussion there too
#
ben_thatmustbeme
but you may be right, just stay here to keep things in one spot
#
snarfed
shrugs
#
snarfed
honestly, i think the biggest counterargument may be that it's not much of a simplification
#
snarfed
currently, the sender specifies the target. with this, the recipient returns the target(s)
#
ben_thatmustbeme
snarfed: the reason target is required is that it might not be a mention you wan to parse at all
#
ben_thatmustbeme
i don't bother even pulling the HTML
#
voxpelli
I don't see any issue with the current behavior – I don't see the problem that's being solved by not specifying a target
#
ben_thatmustbeme
i'll never display it anywhere
#
snarfed
ben_thatmustbeme: that's definitely nice. it's just an optimization though
#
csarven
voxpelli It was meant as a simplification, not raised as an issue on the current.
#
snarfed
but that does count
#
ben_thatmustbeme
snarfed: its also a matter of handling webmentions differently based on target, its a pretty big optimization in the case where you have to hand the webmention off to other code
#
snarfed
there's also the optimization that senders usually know their content better than targets, so they're better suited to find the target urls in a page for a given endpoint
#
voxpelli
csarven: well, if there's no problem with the way it is – and if there's an apparent case where the current way actually provides some advantages – then why "simplify" it if there's no issue with the current approach?
#
ben_thatmustbeme
the final part is for security. if you have, say, a service for webmentions
#
snarfed
so we're down to, the proposal works, it just adds hidden complexities that may outweigh the POST param simplification
#
ben_thatmustbeme
webmention.io, and you refer to someone externally that does not have access to that post, but reply privately to someone who does
#
snarfed
(voxpelli: e.g. tantek would probably argue that simplicity is a good goal for its own sake :P)
#
ben_thatmustbeme
or leave out the externally referenced person
#
ben_thatmustbeme
the service has no clue what key to use to get private data
#
voxpelli
snarfed: sure, but we haven't proven that it's actually simpler yet, less parameters != simpler
#
snarfed
voxpelli: definitely! i'm just putting words in csarven's mouth, since simplicity was his original claim
#
ben_thatmustbeme
snarfed: tantek will argue what tantek will, i wouldn't expect to speak for him, but i would guess this will be, it works well right now arguement
#
voxpelli
sometimes it's simpler to be explicit rather than implicit about things, I rather like that in the current explicit approach
#
snarfed
agreed!
#
snarfed
ben_thatmustbeme: i don't quite follow that yet. afaik services like wm.io can't handle private wms anyway, regardless of whether target is present
chalettu joined the channel
#
snarfed
(unless you assume extra ACL/permission features, in which case we're getting pretty far afield)
#
voxpelli
I would certainly have to change my webmention.herokuapp.com endpoint rather much if I we're to accept target-less pings
#
snarfed
yup, bridgy and wm.io too
#
csarven
One simplification is that the target URL doesn't make the last call. As I understand it, source URL makes the call on the target URL. At the time of submitting the webmention, the target URL is only a "temporary" remark about the relation, nothing more. There is a claim, but it is not verified. The "owner" of the source URL makes the claim and fosters verification.
#
csarven
If the target URL at the time of submission differs than the target URL of the source URL, what then? Which is precisely why the source URL's mention of the target URL matters most.
#
voxpelli
snarfed: ben_thatmustbeme: what's private webmentions? has anyone done such? I think I've missed it if so
#
csarven
Otherwise, any third party can go and make those claims.
#
voxpelli
csarven: if the source page chooses to send a different target URL to the WebMention endpoint than it includes in its own page, then the source page is just acting weirdly?
#
ben_thatmustbeme
voxpelli: there is work on it, private posts you send a standard webmention and the target site has to include a token to get ability to read it
#
snarfed
in practice we haven't seen any use case for wm'ing with target A1 when the source page actually has A2
#
snarfed
returning an error has worked fine so far
#
ben_thatmustbeme
aaronpk: has already done this i believe
#
voxpelli
ben_thatmustbeme: snarfed: I guess a shared endpoint like mine could handle that – with some fund token juggling and such :)
#
voxpelli
s/fund/fun/
#
Loqi
voxpelli meant to say: ben_thatmustbeme: snarfed: I guess a shared endpoint like mine could handle that – with some fun token juggling and such :)
#
snarfed
shudders
#
csarven
In that regard, how many permalinks (and again, I am not talking about a page with a feed on it, but instead an individual entry) have we seen making multiple webmentions?
#
voxpelli
I've been thinking of adding push notifications to the endpoint anyhow so getting private webmentions in there would make it pretty much a plug and play Twitter DM equivilant :)
#
snarfed
csarven: plenty
#
voxpelli
csarven: multiple webmentions from a single page is pretty ordinary – I mean – just look at Twitter for a silo-example – even in those 140 chars there's often very many mentions
#
csarven
Really? Is it documented? e.g., h-entry is a reply to h-entry1 h-entry2... h-entryn?
#
csarven
Wow, I'm not sure how many times I have to repeat myself.
#
csarven
I'm not talking about a friggin' page ;)
#
ben_thatmustbeme
csarven: i make multiple webmentions a lot, (i send two webmentions to brid.gy for syndication a lot) plus some people do multiple in-reply-to
#
ben_thatmustbeme
its all in one h-entry
#
voxpelli
permalink / page– I think we're talking about the same thing, no?
#
snarfed
also, csarven we *are* talking about pages, since pages send wms just like replies/likes/etc :P
#
csarven
ben_thatmustbeme That's fine. Is that a very common thing? How many people doing that?
#
ben_thatmustbeme
csarven: by sending multiple webmentions
interactivist joined the channel
#
ben_thatmustbeme
they usually have to go to different sites entirely anyway
#
csarven
What's the process on that by the way.. do you individually send the source/target pair per POST?
#
snarfed
csarven: yes
#
ben_thatmustbeme
csarven: think of a webmention as a claim to a relationship existing, you claim 1 relationship per POST, webmention endpoint gets data from that URL to verify that specific relationship
#
csarven
voxpelli AFAIK, Twitter doesn't do a reply to multiple tweets.
#
voxpelli
even if all mentions goes to pages/permalinks on the same domain name, they could still have individual endpoints, so one has to look the endpoint up for each post and then individually ping that endpoint after each resolve
#
voxpelli
csarven: WebMentions only does mentions though, replies is an mf2-addition on top of it ;)
#
ben_thatmustbeme
you can @mention as many people as you like (or rather can fit) in twitter
#
ben_thatmustbeme
so yes, it does multiple mentions
#
csarven
Is there a "WebMention" property or is the relation inferred based on looking at the h-* information between source/target?
#
voxpelli
and Twitter does multiple mentions, in my last 5 tweets I have mentioned at least 2 people in each
#
csarven
Lets say in the case of people interested in mf2 in WebMentions.
#
voxpelli
if you find that the target link has an mf2-property when you verify the WebMention, then one can chose to use that to "upgrade" the relationship from a mention to a reply, like, bookmark, RSVP etc
#
csarven
In other words, how does a target know the kind of a relation made/claimed by source?
#
ben_thatmustbeme
waht voxpelli said
#
ben_thatmustbeme
default to mention, learn more from parsing
#
voxpelli
I've chosen to mainly look for like and repost metadata at https://webmention.herokuapp.com/ and constructed the "facepile" out of that
#
voxpelli
(I also parse whether it's a reply or just a mention, but I currently don't really use that data for anything)
#
ben_thatmustbeme
i need to add repost tracking to mine
#
voxpelli
there's quite a few different relation types that people experiment with: http://indiewebcamp.com/interactions
#
csarven
When you say "mention", are you using that as a general concept or is there a property called "mention"?
#
voxpelli
just the general concept of one page including the url of another
#
snarfed
csarven: mostly general concept. a few people use u-mention-of but not many
#
csarven
Looks at /interactions
#
voxpelli
u-mention-of feels a bit redundant to me :/ but probably that question of being explicit vs implicit about things again
#
voxpelli
ben_thatmustbeme: regarding the private posts – would the token be the same kind of token one sends when doing a Micropub call? (but probably with a "read" scope instead of "post"?)
#
voxpelli
indieauth token, right?
#
csarven
So, I take it that WebMention endpoints (interested in mf2 or alike) would be interested in keeping a local copy of the vocabulary used in practice at /interactions
#
snarfed
voxpelli: afaik yes. aaronpk and ben_thatmustbeme et al discussed this at length yesterday, and wikified a bit. probably worth reading logs
#
voxpelli
snarfed: so hard to keep up on everything :P
#
snarfed
csarven: often yes
#
ben_thatmustbeme
voxpelli: could be anything, but thats the simplest implementation, there is discussion on how to get that token across best still
parzzix joined the channel
#
ben_thatmustbeme
i have stopped trying
#
voxpelli
ben_thatmustbeme: got a wiki page?
#
ben_thatmustbeme
it gets too much sometimes
#
voxpelli
I just made things a bit more complicated by creating a Micropub endpoint in addition to my WebMentions endpoint – so now I have to keep up with everything regarding both :P
#
snarfed
(csarven: although "vocabulary" is a bit strong, since most people only support like, repost, and maybe rsvp)
#
ben_thatmustbeme
what is private-webmention? i think that has a some of the data voxpelli, i need to get up to date on some of the discussion myself
#
Loqi
private-webmention is a suggested protocol for combining microformats2, Webmention, and IndieAuth to convey private posts between two people or between groups of people https://indiewebcamp.com/private-webmention
#
csarven
snarfed A list of terms :) Small dictionary.. a small definition-list.. just semantics :)
#
voxpelli
ben_thatmustbeme: cool, thanks, then I have something to bookmark as a reference for when I get time to look at it :) I think eg. salmentions will have a higher priority
#
ben_thatmustbeme
voxpelli: same here
wolftune, jciv, shiflett and chreekat joined the channel
#
kylewm
aaronpk: is it possible to differentiate a webmention that was proxied from webmention.io's pingback handler from a real one? it looks like the U-A is just "Ruby"
#
@jgarber
@kissane @heywren This is a part of what the #indieweb movement is trying to accomplish, but it’s early days and still very technical.
(twitter.com/_/status/621336131116900352)
Pierre-O joined the channel
#
@lackaff
The death of the Web, the rise of the Stream, & the "jealous gaze" of social media: https://medium.com/matter/the-web-we-have-to-save-2eb1fe15a426 #indieweb
(twitter.com/_/status/621336253665902592)
#
@jgarber
@kissane @heywren Events like IndieWebCamp are still niche, but the desire to create tools for all is there, I think. http://indiewebcamp.com
(twitter.com/_/status/621336449632354304)
edpw and j12t joined the channel
#
@voxpelli
@pragmaticcode Have you looked at adding Micropub to Linky? An open API effort by the @indiewebcamp community: http://indiewebcamp.com/micropub
(twitter.com/_/status/621337468701396993)
snarfed and KevinMarks joined the channel
#
aaronpk
reads scrollback
#
aaronpk
wow I can't believe anyone is suggesting that the webmention response would be delayed until the endpoint parsed the source page
#
aaronpk
looks like the webmention discussion ended with no resolution?
#
aaronpk
kylewm: you want some indication that webmention.io is sending you a forwarded pingback? sounds like a good issue to open
snarfed joined the channel
#
aaronpk
I could add something to the user agent, or add an HTTP header, whatever
#
snarfed
aaronpk: csarven later modified his proposal to not require returning the target URL, so async would be ok
#
snarfed
i think the conclusion was that it's technically doable, but probably adds more implementation complexity than it removes
#
aaronpk
the proposal was basicaly to drop the "target" parameter, right?
#
snarfed
so recipients would handle any links in the source doc that they can
#
csarven
That is pretty much the case right now any way.
#
aaronpk
so in that case, the webmention becomes more of a "hey you might be interested in this page" notification
#
csarven
That is the case right now as well.
#
aaronpk
no, right now it's "hey this page linked to this other page, go verify it"
#
csarven
Practically the same thing, but okay, we can go with that description.
#
csarven
Everything remains the same. The difference is again, a simplification by dropping the target URL at submission time.
#
voxpelli
(people sometimes uses it to link to their own posts as well, when eg. doing a multi-part blog post or continuing one some othe rline of thought they had previously posted)
#
csarven
It is not needed as one would have to verify in any case.
#
voxpelli
it communicates intent and context
#
aaronparecki.com
edited /Webmention (+29) "/* verifying private webmentions */"
(view diff)
#
csarven
Gar. I have to run. I'll have to read/reply later.
KevinMarks__ joined the channel
#
@HongPong
a high point for my experience lurking at MIT when @timberners_lee made off with a burrito from #indiewebcamp https://twitter.com/business/status/621338668305883136
(twitter.com/_/status/621340790296588288)
#
csarven
snarfed I don't recall proposing what you just said.
#
aaronpk
I guess I just don't see why it's simpler at all. sure it's one less parameter in the post request, but the sender already has that value anyway
#
aaronpk
because that's the URL they found the webmention endpoint at
snarfed joined the channel
#
csarven
Redundant as I see it.
#
snarfed
csarven: apologies if i misquoted you. it probably does need that change though, otherwise you can't do async, which is pretty important
#
aaronpk
yeah there's no way i'm gonna do webmention processing synchronously
loic_m joined the channel
#
aaronpk
so i'm gonna continue this thread as if the proposal allows async procesing
#
csarven
snarfed No worries. I may have missed that suggestion earlier.
#
aaronpk
here's why it's not redundant, and is actually a good idea to include in the webmention request
#
aaronpk
when I get a webmention request, I need to verify that it's valid, and I also optionally parse the page for microformats to "upgrade" it to a like/repost/whatever
#
aaronpk
when the target parameter is in the request, the initial verification is easy, because I know the pattern on the page I'm looking for
#
csarven
Your "verification" is based on the POST? You don't check the actual source URL to physically see whether it is pointing to target URL?
#
aaronpk
what no
#
aaronpk
of course i check the source that's what i'mt alking about
#
aaronpk
when I go fetch the source HTML, I'm looking for the specific string that was in the "target" parameter
#
csarven
Then it is the source that tells you what it is targeting in reality, not the information in the pOST>
#
aaronpk
listen to what i'm saying
#
csarven
You may be using the POST for the purpose of other htings.
#
csarven
OK
#
csarven
BBL...
#
aaronpk
with the target parameter, I am looking for a specific pattern. without the target parameter, I don't know exactly what I'm looking for
#
snarfed
aaronpk: right. you'd have to look for all urls (or domains, etc) that your endpoint handles
#
snarfed
this is the added implementation complexity i alluded to, and that we discussed earlier
#
aaronpk
therefore it's easier for me to handle the webmention request if the target parameter is included
#
voxpelli
and self-mentions would be impossible as _all_ links to ones own blog would always be mentioned on each and every self-ping
#
aaronpk
and since it's no extra burden on the sender to include it, and makes it easier for the receiver, it's a good idea to do it
#
aaronpk
does that sound like a reasonable justification?
#
aaronpk
I'm going to add that to the wiki as an FAQ, and hopefully we can close this discussion
#
voxpelli
I like it
tvn and cweiske joined the channel
#
aaronparecki.com
edited /Webmention (+943) "/* FAQ */ why is target required"
(view diff)
KevinMarks, gRegorLove, GWG and tantek joined the channel
#
tantek
aaronpk - definitely add csarven's simplification brainstorm and follow-up to the wiki. Since he came up with it, it's likely someone else will come up with it again in the future, and having a good explanation for why it's overall not simpler (sender+receiver overall complexity) will help.
#
Loqi
tantek: ben_thatmustbeme left you a message on 7/14 at 1:55pm: for person B support, Person-tag is written but completely untested and i need to set up receiving salmentions, but then i have to rewrite all of my notifications and mention processing stuff sometime soon http://indiewebcamp.com/irc/2015-07-14/line/1436907317741
#
Loqi
tantek: ben_thatmustbeme left you a message on 7/14 at 1:55pm: Cool, can't wait to get Web Notifications cross browser http://indiewebcamp.com/irc/2015-07-14/line/1436907345426
#
ben_thatmustbeme
'self-mentions would be impossible' really sums it up for me
#
tantek
my addition: don't forget that the receiver *also* has to do redirect handling on the target URL, which is MUCH easier/simpler with *one* specified target URL, whereas having to do a HEAD request on every URL in the source to see if it *might* link to the target "somewhere" is A LOT more complexity and work.
#
aaronpk
tantek: good point about following redirects!
#
tantek
in addition, for any kind of "sending" type protocol (which webmention is), it's much better to balance the work between sender and receiver, than to fully burden the receiver (which e.g. SMTP does, and was clearly a mistake, spam/forgeries etc.)
#
aaronpk
already added a bunch to the wiki, will add that
#
tantek
thank you!
j12t joined the channel
#
tantek
this kind of falsification of a potential simplification documentation is essential for longevity of design
#
tantek
it helps establish that we (collectively as a community) have considered nearby possibilities/alternatives
#
ben_thatmustbeme
oh yeah, don't know how we all missed that
#
ben_thatmustbeme
the redirects part
#
tantek
yeah it's kind of a big burden
#
ben_thatmustbeme
Woah, tantek, i just noticed that the <title> tag to your posts contain the entire content of the post
#
ben_thatmustbeme
i'm guessing thats because the post has no title and thus from MF2 parsing it uses the content
#
aaronparecki.com
edited /Webmention (+251) "/* Why is the target URL a required parameter */ add note about following redirects"
(view diff)
#
tantek
ben_thatmustbeme: hah - did you see the screenshots kevinmarks posted?
#
ben_thatmustbeme
hah, he noticed it the same way i did
#
ben_thatmustbeme
i happened to hover over the tab and said "what the hell was that?"
#
tantek
also the reason it's like that is a combination of legacy and laziness
#
tantek
originally I published that way as a hint to consuming applications/code that it was a note - that is title==content
#
tantek
however I don't think that anyone depends on / uses that anymore
#
tantek
and frankly auto_link (of the content but not title obv) breaks that too
#
tantek
so now it's a combination of design problem (how to best shorten the synthetic title for notes?) and laziness (haven't implemented even a dumb shortening of synthetic title)
#
tantek
design/algorithm suggestions welcome. best I have so far is first sentence and/or ellipse to 70 char length at most.
#
ben_thatmustbeme
i think thats a generally reasonable method to use
dpg1, uhhyeahbret and Sebastien-L joined the channel
#
tantek
ben_thatmustbeme: how do you generate the <title> of your note posts?
#
aaronpk
Jeena: ^^
#
tantek
oh boy, location change day of?
#
aaronpk
well it's the same location as last week
#
tantek
maybe consider putting up a physical sign at the previous location?
#
tantek
so anyone that errantly shows up there (because they already copy/pasted it into their calendar) gets the fwd?
#
tantek
also indieevent and all POSSE copies (FB, calagator)
#
tantek
hey that reminds me
#
aaronpk
i don't think i have a facebook event for this week yet
#
tantek
aaronpk - when you update the location on your indieevent post, do you send webmentions to all the RSVPs and invitations?
#
tantek
(did I just make up a new update protocol?)
rknLA joined the channel
#
Jeena
hehe neat aaronpk :D
#
tantek
and of course that should result in notifications for those that RSVPd or were invited
#
tantek
"aaronparecki.com changed the location of the event Homebrew Website Club Meetup Portland"
#
Jeena
I still can't automatically RSVP with my website
#
ben_thatmustbeme
tantek: I don't, the site generically starts with Ben Roberts, and then if there is a title to whatever post you are viewing it adds that
#
aaronpk
isn't that what a salmention is?
#
tantek
aaronpk: I don't think so, because your indie event is not in-reply-to anything
#
aaronpk
or did we not specify that webmentions should be re-sent to comments
#
tantek
and you're not passing along an external change, but rather YOU made a change to your post
#
aaronpk
if you send webmentions for any updates you make to your post, you'd be doing this already by default
#
jeena.net
edited /events/2015-07-15-homebrew-website-club (+32) "/* RSVP */ Portland Jeena"
(view diff)
#
tantek
aaronpk - I'm not sure people think about sending webmentions to links they didn't create themselves (e.g. links created automatically by linking back to comments on a post)
#
tantek
the only variant of this that I've proposed before is reply-context updates
#
ben_thatmustbeme
salmentions go upstream, this is a message in a bottle floating downstream
#
aaronpk
hm, I suppose it's worth calling out explicitly
#
aaronpk
even though the link is on the page, it is likely not in the internal canonical version of the page
#
Jeena
yeah, I don't do that yet, I only webmention the links in my own content but that is a good point
#
tantek
that is, when you receive a comment/like/repost, and then you update your original post, you should send a webmention to the comment/like/repost wherein it can notice that the thing they are a response to has been updated, and update their reply-context / like-summary/context / or perhaps entire repost?
#
tantek
the event case is a bit different still
#
aaronpk
for example the contents of my posts are in a file, but the comments live elsewhere in the filesystem. i actually internally render the HTML of just my post and send webmentions to those links, so it is additional code for me to send webmentions to comment URLs
#
tantek
aaronpk - precisely
#
tantek
the event case is different because invitations
#
tantek
that is, invitations are content *you* added to your indie event, not from an external source
#
tantek
and they require webmentions to a homepage, not a post permalink
#
aaronpk
yes but RSVPs may be unsolicited replies just like comments
#
tantek
correct
#
Zegnat
I am not sure I would even expect a webmention coming downstream, as the original post doesn’t actually “mention” me
#
aaronpk
for anyone who RSVPd to be able to render notification text like "aaronparecki.com changed the location of an event" it means they need to store the event at the time they RSVPd so they can tell what changed
#
tantek
Zegnat - it does once it links back to you!
#
tantek
aaronpk - yes!
#
tantek
that should be in the reply-context of the RSVP!
#
tantek
(now we're getting somewhere)
#
aaronpk
right!
#
Zegnat
tantek, does that mean that, for every comment I add to my page, I should send a webmention to all other comments as well? That sounds like people would soon get flooded with webmentions
#
aaronpk
good, so there are 2 reasons to store the data
#
aaronpk
sending webmentions to everyone who commented when a new comment comes in would enable the notification text "someone commented on a post you commented on"
#
Zegnat
yes, but also means that if all of them parse the comment thread and update theirs and do the same thing you will be notified of the same comment thread on X different websites
#
voxpelli
aaronpk: that would be enabled by subscribing to new comments through PuSH as well? (not sure if you've already mentioned that, should read scrollback)
#
aaronpk
oh yeah PuSH is another way of doing that
tilgovi joined the channel
#
voxpelli
I'm thinking WebMentions for targeted pings and PuSH for broadcasted "pings"
edpw joined the channel
#
tantek
Zegnat - yes! FB does this with notifications!
#
tantek
"someone commented on a post you are following"
#
tantek
once you have commented on a post you get that ^^^ notification for all subsequent posts
#
tantek
er, maybe more like "someone commented on a post you commented on"
#
tantek
Zegnat good catch!
#
tantek
reads aaronpk already saying the same thing :P
#
tantek
what voxpelli said
#
Zegnat
"someone commented on a post you commented on" but wouldn't you get that notification for every site that mirrors the comment thread? And all those sites have different permalinks, so can’t easily be filtered out? Maybe I am overthinking it and we should wait for implementation
#
tantek
wow if we actually end up spec'ing all the different notification variants that FB / IG / Twitter etc. do, but with federated webmentions, I think we win.
#
tantek
Zegnat - no you only get it directly from the site you commented on
#
tantek
that's part of the elegance of the salmentions design!
#
voxpelli
But what if one wants to unsubscribe/unfollow? Removing a PuSH-subscription is easy, removing a WebMention "subscription" is impossible(?)
#
Zegnat
no. if my site has threaded comments and with salmentions let me add deeper levels it will grow long with comments from people who never actually replied directly to me. But if I have to send a webmention downstream to every link on my page (that includes replies-to-replies), and everyone else does that too, your 3 levels deep comment can easily get a lot of webmentions. One from each site that mirrors the discussion
#
Zegnat
or should the reverse-salmentions be specified to *only* send to direct replies / RSVPs? That would solve it, but I don’t think anyone mentioned that yet
#
Loqi
slack/snarfed: voxpelli: probably the right way to handle that is on your own site, ie choosing which kind of wms you send yourself push (if whatever) notifications for
#
voxpelli
Zegnat: I would say the downstream ones would have to subscribe to the upstream parts over PuSH if they want to get updates, WebMentions / Salmentions should only go upstream imho, PuSH is what flows downstream
#
Loqi
slack/snarfed: ie only loosely couple webmentions (plumbing) with notifications (UX)
#
voxpelli
feels preferable to get the fat pings of PuSH as well if one starts to get very many interactions to ones post
#
voxpelli
a conference with 500 attendees would be much better of notifying people that has subscribed to updates rather than just sending WebMention pings to all those who has RSVP:ed and have all of them at once verify the WebMention
#
@garethjordan
The Web We Have to Save - The Web We Have to Save https://medium.com/matter/the-web-we-have-to-save-2eb1fe15a426 via DuckDuckGo for Android #IndieWeb #OpenSource #FOSS
(twitter.com/_/status/621361703381405696)
#
voxpelli
or am I missing something?
#
tantek
voxpelli: it's a good question (500 webmentions vs. PuSH)
#
aaronpk
I think I like the PuSH for downstream updates, webmention for upstream comments thing
j12t and snarfed joined the channel
snarfed1 joined the channel
#
rhiaro
catches up on last 4 hours of logs
#
tantek
some goodtimes in there rhiaro
evalica, csarven and tantek joined the channel
#
tantek
!tell benwerd note comments markup updated slightly with u-comment instead of p-comment, see https://indiewebcamp.com/comments-presentation#How_to_markup for update and reasons!
#
Loqi
Ok, I'll tell them that when I see them next
JarOfGreen joined the channel
#
bret
what other "type"s can be used with the <script> tag?
#
cweiske
vbscript
#
aaronpk
some people cram templates or data into pages using the script tag too
#
bret
oh its also for accessing ES6 features in firefox
#
tantek
bret - see also Microsoft's failed XML Data Islands and Google's reinvention of it with JSONLD Data Islands (which will likely fail in 5-10 years like XML Data Islands did)
#
bret
i was just ¯\_(ツ)_/¯ about it... i always put JS in it
#
tantek
(from 2011)
wolftune joined the channel
#
bret
yeah i remember that tweet
#
bret
they want to belive
#
tantek.com
edited /events/2015-07-15-homebrew-website-club (+214) "/* What */ Topics for tonight, IWC 2015 review (SWAT0!!), "The Web We Have To Save""
(view diff)
#
cweiske
ha. xml data islands
#
cweiske
I actually used that at the time
#
cweiske
but I also spit out XML with XSL stylesheets
#
cweiske
and didn't care about browsers that did not ship with XSLT
#
GWG
So, I tried to add query support to the WordPress Micropub plugin and thought I was doing something wrong till I realized that the endpoint is a query string and thus is an issue for adding a query string to
#
GWG
Trying to figure out how to address it
#
aaronpk
oh funny. well clients should be able to add to the query string
#
GWG
aaronpk: I was trying with Quill
#
aaronpk
sounds like a bug in quill then ;)
#
GWG
That is what I was wondering. I think it would be an issue for most clients
#
GWG
I filed an issue.
#
GWG
Of course, my query code is not in the stable version yet.
#
GWG
I sort of wanted to test it before the PR. Better try the other clients
#
aaronpk
what are you adding?
wolftune joined the channel
#
GWG
aaronpk: Syndicate-to was this morning
#
GWG
Some misc fixes. Support for my plugins and post formats.
#
tantek.com
edited /events/2015-07-15-homebrew-website-club (+29) "/* Topics */ feel free to add more"
(view diff)
#
GWG
Basically adding more from the specs while making it useful for me
lukebrooker, benwerd and awolf joined the channel
#
csarven
aaronpk The current design is arbitrary because it doesn't specify how to handle redundancy or errors either. It is not "extra burden" to add/leave the target parameter if you are the one that's coming up with the definition of the complexity. The point is that, it is *safer* to ignore the target URL in submission, because it becomes obsolete the second you look into the source URL - which is something you have to do any way! Your
#
csarven
concern on what to look for is as well an arbitrary concern. source URL is simply the atomic item (on a page) which made the reference to the target URL. It is quite trivial to figure out what to look for especially if one can count the number of possible properties under IWC/interactions.
#
aaronpk
csarven: did you not see my comment about looking for the specific URL vs looking for all URLs *after* following redirects?!
#
aaronpk
for example, good luck convincing anyone to write code that accepts your "source" parameter where they then have to look for all URLs on the page, follow redirects on all, and look for any that match their site
#
csarven
What "all URLs" are you referring to? Again, Webmention doesn't specify in any way how the receiver should handle/parse the response, so why are you imposing that is somehow an issue?
#
csarven
If you do give the instructions and people go out of their way, okay, then we can discuss.
#
aaronpk
webmention requires that the receiver verify the link exists
#
csarven
But Webmention is not giving those instructions whatsoever
#
@voxpelli
Updated a lot of dependencies + moved from Node 0.10 to iojs on my WebMention endpoint: https://webmention.herokuapp.com/ Please ping me if it acts up
(twitter.com/_/status/621388096022687744)
#
aaronpk
what are you talking about
#
aaronpk
webmention verification is a ting
#
csarven
I'm well aware of that. I have stressed that point 50 times now.
#
aaronpk
nothing about webmention verification requires microformats, sure, but it does say to go look at the source URL and verify the link to your site exists
#
csarven
Webmention endpoint tries to verify x relation y. Period. x and y can be anywhere.
#
csarven
I never said anything about mf. Why are you bringing thta up even?
#
aaronpk
in case you were confused? i'm not sure
#
csarven
Minimize your assumptions. Just a suggestion.
#
aaronpk
webmention says the receiver goes and fetches the source URL and looks for the target link
#
aaronpk
if the target link is not specified, the receiver has to look at ALL URLS on the page, follow redirects on all, and decide which are targets
#
csarven
"Webmention endpoint tries to verify x relation y. Period. x and y can be anywhere. " is significantly different than "go look at the source URL and verify the link to your site exists"
#
aaronpk
i don't know where you got "Webmention endpoint tries to verify x relation y" from
#
csarven
perhaps you are confused? :)
#
@voxpelli
@maxogden thanks! lots of interesting things happening in the @indiewebcamp community, check eg. https://www.brid.gy/ and such as well :)
(twitter.com/_/status/621389077322039297)
#
aaronpk
and i don't know what you mean by "can be anywhere"
#
csarven
I'm looking at the README for WM.
#
csarven
anywhere on the Web
#
csarven
under any domain.
#
csarven
controlled by any authority.
#
aaronpk
"relation" does not appear on either webmention page
#
aaronpk
so i still don't know what you're talking about
#
csarven
Argh.. I'll use terms that you can understand.
#
voxpelli
relation == mention I assume
#
csarven
How about a "property"
#
csarven
Yes a "mention" will do
#
csarven
The "property" is not specified by Webmention.
#
csarven
Nor the data model.
#
csarven
So.. that whole thing is completley open for the consumer / webpoint endpoint.
#
csarven
s/webpoint/webmention
#
Loqi
csarven meant to say: So.. that whole thing is completley open for the consumer / webmention endpoint.
#
aaronpk
csarven: out of curiosity, have you implemented either sending or receiving webmentions yet?
#
csarven
The point is that, there is some sort of a statement made by source URL about target URL with a particular kind of a relation. That relation ... is what you conceptually call a "mention". There are some properties for it documented at http://indiewebcamp.com/interactions as I'm told.
#
csarven
aaronpk Not yet. Is that relevant to the discussion?
#
aaronpk
potentially yes. because arguments not based on implemention experience tend to be less useful.
#
snarfed1
aaronpk++
#
csarven
heh
#
Loqi
aaronpk has 886 karma
#
csarven
snarfed1++
#
Loqi
snarfed1 has 1 karma
#
csarven
snarfed++
#
Loqi
snarfed has 114 karma
#
csarven
Karmas all around.
#
voxpelli
what is selfdogfooding?
#
Loqi
selfdogfood is a stronger form of dogfooding, that is, using your own creations on your own personal site that you depend on, as an aspect of your primary online identity, day to day — if you're not willing to use your creation on your own primary personal website, why should anyone else use it on their primary personal website? https://indiewebcamp.com/selfdogfooding
#
csarven
voxpelli++
#
Loqi
slow down!
#
gRegorLove
/interactions is about extracting meaning from the mf2 on the source page, not directly related to the webmention itself
#
gRegorLove
As I understand it
#
csarven
gRegorLove Right.
#
csarven
But it is the only tangible thing we can point at that's used along with Webmention as I've gathered here.
#
voxpelli
csarven: most WebMention implementations doesn't care what type of a mention a mention is
#
voxpelli
most IndieWeb-implementations that is
#
snarfed
csarven: mostly, but not entirely. other small webmention uses include salmentions, bridgy publish triggering (https://www.brid.gy/about#publish), etc
#
voxpelli
some endpoints use it as a way to progressively enhance mentions of known interaction types, but all that I know of falls back to a simple basic mention presentation on unknown types
#
csarven
aaronpk I find it amusing that you had to resort to "have you implemented Webmention" yet.
#
csarven
And with snarfed and voxpelli onboard, I feel ganged up with all the karmas and citations to the Wikis fly aorund.
#
aaronpk
csarven: no, you are just completely ignoring my arguments
#
csarven
You are not at all addressing what I'm stating.
#
gRegorLove
Right, voxpelli. In my implementation I parse for in-reply-to mf2 and if so, extract the plaintext content as a comment. Otherwise I just show "[User] mentioned this"
#
csarven
The decisions you are making right now are completely arbitrary.
#
aaronpk
not arbitrary, they are based on implmentation experience
#
csarven
Right, and I'm suggesting a simplication because wha tyou have is redundant.
#
voxpelli
gRegorLove: interesting! I always show the "content" if it isn't a repost or a like
#
aaronpk
i thought i pointed out why it's not redundanty
#
csarven
Whatever
#
gRegorLove
Er, I guess I do parse /likes as well.
#
aaronpk
webmention is a protocol that says one page linked to another page
#
csarven
++ to you for editing a wiki page
j12t joined the channel
#
csarven
Should I go and create a bunch of URLs with my argument?
#
voxpelli
gRegorLove: so I actually don't make a difference between a reply and a mention currently
#
csarven
Would an URL to the IRC chat log suffice?
#
aaronpk
webmention is not a protocol for saying one person liked a post by another person
#
aaronpk
i put it on the wiki because it will likely be asked again, so thanks for asking the question in the first place
#
voxpelli
csarven: well, the flow that often happens in discussions here are that people implement things, documents experiences and iterates on that over and over – hence the general strong focus on selfdogfeeding
#
gRegorLove
voxpelli: I think the main reason I'm doing that is in case a blog post links to me, I don't want to display the entire blog post.
#
csarven
heh
#
ben_thatmustbeme
looks in. Is this arguement still going on?
#
csarven
With what.. like 5 people implementing?
#
gRegorLove
I think adactio doesn't make that distinction either. I wrote an article linking to him and the whole post showed up on his page, heh.
#
voxpelli
gRegorLove: right, I think I limit the length of the post content and just shows a "..." if its too long
#
csarven
Hey, you guys go ahead. I'm going to excuse myself. It seems like IWC is allergic to new ideas or their own ideas being challenged.
#
aaronpk
csarven: like i said, thanks for bringing it up, but i disagree with the assertion you made and i documented why
#
csarven
I have yet to hear a sound argument against the simplification meanwhile not changing the fundmanetals of Webmention
#
csarven
aaronpk I appreciate your acknowledgement and response. KUTGW
#
ben_thatmustbeme
csarven: " It seems like IWC is allergic to new ideas or their own ideas being challenged."? seems to be getting a bit unnecessary
#
ben_thatmustbeme
usually after like 5 or 6 people stop putting their name on the wiki
#
ben_thatmustbeme
at least from what i have observed
#
ben_thatmustbeme
that page does have a lot of implementations listed
#
csarven
re: http://indiewebcamp.com/webmention#Why_is_the_target_URL_a_required_parameter .. receiver also doesn't know which "mention" to look for. however, it makes a decision regardless on what it is interested in. So, why is that on one hand the "easier" argument is made when it is not at all essential to the architecture of Webmention, and on another somehow "what to look for" is this rocket science. Wakeup call: the receiver is maki
#
csarven
ng all the decisions in the end - Webmention doesn't specify anything any way other than saying that, go and look at the source URL to see if there is a relation to a target URL that you want.
#
aaronpk
okay I think I see what you're getting at
#
csarven
So, that is arbitrary design IMHO
#
csarven
I'm not saying it is bad.
#
aaronpk
i'll tell you why i'ts not arbitrary, one sec
#
csarven
I'll even buy the argument on "easier".
#
csarven
ben_thatmustbeme Sorry, you are right. A bit unnecessary :) I'd like to retract that.. Would Loqi allow? =)
#
ben_thatmustbeme
he seems to smile upon that idea
#
aaronpk
from start to finish, (receiving initial POST request to displaying a "like" on the page) you are correct that the receiver has to look at the source page and figure it out from there, and the target parameter doesn't help with figuring out the *meaning* of the webmention
#
aaronpk
however, the semantics of the like/repost/comment/whatever is secondary to the webmention itself, since a page with no microformats can still be useful to displaya on the target page
#
csarven
Is the latter a simple x hyperlinks to b?
#
aaronpk
because not all pages contain microformats, receivers have to first look for a plain link to their page on the source page
tilgovi joined the channel
#
aaronpk
then they optionally upgrade that from a generic "mention" or "links to" to something more specific like "commented on" or "liked"
#
aaronpk
it gets better:
#
aaronpk
because there is no authentication, this is an easy target for spam. a webmention endpoint needs to be able to reject bogus requests as quickly as possible.
#
aaronpk
which is why the simplest possible initial check is important, before doing things like microformats parsing and inspecting that data structure
#
csarven
Why not give the complete statement to the Webmention endpoint?
#
csarven
e.g., h-entryX u-in-reply-to h-entryY (only an example)
#
aaronpk
that is a potentially useful addition, but a totally different idea
#
ben_thatmustbeme
because you can't trust it
#
csarven
The endpoint can at least know whether to decide to work with mf2 or not.
#
aaronpk
you can't trust it, but it would help you know where to look on the page
#
csarven
I agree that you can't ultimately trust it unless you check the source URL.
#
ben_thatmustbeme
indeed. though for that i would suggest #fragment should be used really
#
csarven
But given the complete statement will tell the endpoint precisely WHICH statement to verify
#
ben_thatmustbeme
plus you then have to question what to do when they tell you its in-reply-to, but its not, its actually just a mention, etc
#
aaronpk
if you're going to consider adding something to the publishing and consuming of webmention, you have to consider how difficult it is for both parties to deal with
#
csarven
Because.. if your h-entry states "in-reply-to" as well as "likes" to my h-entry, well, that's something the current Webmention endpoint can't precisely figure out. It will probably take both.
#
voxpelli
as far as I see it, as long as there's a link at all the verification is successful – everything else is just about presentation and/or filtering of content – that's step 2
#
ben_thatmustbeme
csarven: i treat it as both, i store as a reply and a like
#
aaronpk
voxpelli: yes. i think csarven is implying that since most of the time the webmention is also communicating some other semantics, why not optimize for that? (is that correct csarven?)
#
voxpelli
csarven: it's not a challenge that WebMention's is really conserned with – that's a general challenge you have when presenting an mf2 post – should you emphasize the "like"-part or the "reply" part, chose one or both?
#
ben_thatmustbeme
at least that the theory, not tested since people really don't do it
#
csarven
aaronpk Right. Either simplify and let the endpoint figure it out all (since no real instructions are given but everything is left to the discretion of the target), or be very precise about what you want the webmention endpoint to know (i.e., to look further into)
tilgovi joined the channel
#
csarven
As an added bonus, the endpoint can even decide what to do without even making a single GET request.
#
ben_thatmustbeme
a better way to think about it is this, webmention endpoints all have MF2 processing plugins right now, WM endpoint just verifies that the target URL is at the specified page, and that this URL is of your domain (or redirects to it if you follow it) . The rest is processing done by your MF2 side of things, but this could be swapped for other things
#
csarven
1. Is it mf2? Yes 2. Is it an "in-reply-to"? No. Then I don't want it.
#
voxpelli
I'm thinking of adding additional processing as well, like Open Graph data
#
ben_thatmustbeme
csarven: unless they sent it wrong as like instead of in-reply-to. its a reply you want but now you aren't checking it.
#
csarven
True. But, then again, life is hard :)
#
ben_thatmustbeme
putting mf2 class in to WM sort of breaks seperation of concerns for me
#
csarven
It doesn't have to be an mf2
#
voxpelli
so far no implementations has felt a need for any such extra verification of the "mention-type"
#
csarven
That's left for open
#
csarven
source URL, relation URL, target URL
#
voxpelli
ben_thatmustbeme: +1 on the separation of concerns
#
ben_thatmustbeme
hmm, almost sounds like an extension
#
ben_thatmustbeme
just like vouch
#
ben_thatmustbeme
also, at that point csarven, its really just an optimization
#
voxpelli
an optimization that no current implementation has really felt a need for :/
#
ben_thatmustbeme
you can send it, but chances are people will ignore it, and not send it either
#
voxpelli
I'm rather liberal in what I accept than starting to verify that the mf2 data matches the pinged data – as long as there's any link at all I'm fine, then someone isn't faking something
#
csarven
If the Webmention endpoint sees the relationURL, it knows what to do with it - it may not even want to bother with the HTTP GETing the source URL for the verification.
#
ben_thatmustbeme
whats that quote about being liberal in what you accept and... something
#
aaronpk
right, this is where it goes back to implementation experience. so far nobody has indicated they've had any trouble with the way it currently works
#
kylewm
what is Postel's law?
#
ben_thatmustbeme
csarven: i could point you back in to the conversation earlier where you were arguing thats just an optization, not necessary
#
ben_thatmustbeme
of course you were talking about target= at the time
#
voxpelli
I think much of this boils down to: any endpoint can choose to support or ignore any parameters they want and if that shows to deliver a better overall experience for everyone involved then other implementations will likely follow
#
csarven
side-effect. Not the primary reason to have the relation URL
#
gRegorLove
pokes Loqi
#
Loqi
goes squish.
#
kylewm
Loqi++
#
Loqi
Loqi has 357 karma
#
ben_thatmustbeme
hahaha, Loqi
petermolnar joined the channel
#
petermolnar
good evening, everyone
#
loqi.me
created /Postels_Law (+25) "prompted by gRegorLove and dfn added by gRegorLove"
(view diff)
#
petermolnar
a-not-too-glorfindel
#
petermolnar
s/a-not-too-glorfindel/what the hell happened to my keyboard?/
#
Loqi
petermolnar meant to say: what the hell happened to my keyboard?
#
voxpelli
looks like the twitter bot that mentions everything indieweb and webmentions doesn't really care about micropub stuff, what's up with that aaronpk? :)
#
voxpelli
means one can be all stealthy with ones micropub discussions
#
aaronpk
too many references to actual pubs for a "micropub" search to work
#
aaronpk
tho a search for micropub -beer -scotland might work
#
voxpelli
ah, I see
#
voxpelli
regarding https://twitter.com/pragmaticcode/status/621400788221472768 – has there been any iOS IndieAuth / Micropub development?
#
@pragmaticcode
@voxpelli didn't know Micropub, looks cool but not sure for adding it to Linky in a near future. I keep it in the user's suggestions list.
(twitter.com/_/status/621400788221472768)
#
aaronpk
i have a stupid simple ios app that does
#
cweiske
ben_thatmustbeme, my webmention endpoint does not have MF2 processing
#
aaronpk
ben_thatmustbeme made a real mobile app
#
voxpelli
aaronpk: thinking that anything that lowers the threshold would be good to be able to point someone like that to
#
cweiske
so the "all" was a bit exaggerated
#
aaronpk
i actually cheated with mine and it uses server-side code to do the indieauth login process
#
ben_thatmustbeme
aaronpk: don't you use micropub with your pushup app thing?
#
aaronpk
yeah that's the one
jciv joined the channel
#
Jeena
I need to implement handling of images before I can make a android client which will be able to post offline (it caches it and posts it when there is internet with the right date, etc.)
eschnou joined the channel
#
ben_thatmustbeme
cweiske: it doesn't? interesting. what do you do with webmentions?
#
cweiske
store them in a database and send mails to me
#
cweiske
in the future i might implement WM display on the website
#
cweiske
but that's not high on my priority list
#
cweiske
getting ben not to cry anymore is #1 right now
#
cweiske
another ben :)
#
Jeena
aaronpk, https://indiewebcamp.com/Micropub mentiones mp-syndicate-to is this the preferred way or is it just an old artifact?
#
ben_thatmustbeme
cweiske: too many bens
#
ben_thatmustbeme
thats why ben.thatmustbe.us exists :P
#
ben_thatmustbeme
considers just searching for every person named ben with a personal blog he can find and just adding them all
#
Jeena
ben_thatmustbeme, you have 4 rss links, two times to rss and two times to all content, what is the difference?
#
@voxpelli
@jaredsinclair Thinking a lib that fixes the UI-dance with web redirects which one can create a plugin for – want for http://indiewebcamp.com/IndieAuthProtocol
(twitter.com/_/status/621404990175903744)
#
ben_thatmustbeme
i don't have rss at all on my site?
#
ben_thatmustbeme
ben.thatmustbe.us randomly picks a ben's website and redirects
#
Jeena
I just clicked on a ben at whithknown
#
Jeena
haha ok
#
ben_thatmustbeme
indie.thatmustbe.us randomly picks from everyone on irc-people
#
ben_thatmustbeme
well of those that loaded, i trimmed a few
#
Jeena
I was confused because I just clicked on that link the second time and there was a different site
#
ben_thatmustbeme
or rather those on the page at the time i polled it, i need to fix that to acutally poll over and over
#
ben_thatmustbeme
indieweb roulette
#
ben_thatmustbeme
doesn't remember who coined that term for it
#
Jeena
hehe ok, now I can't find that site anymore :D
Deledrius joined the channel
#
ben_thatmustbeme
thost are the ones on the ben list
#
Jeena
ok it was the mlncn one
#
ben_thatmustbeme
Jeena: i'm guessing that all of Known then
#
voxpelli
Thinking some more on iOS, IndieAuth and MicroPub – the "complication" there would probably be that one would want to authenticate against Twitter with the native authentication rather than the web one
#
ben_thatmustbeme
they ahve different rel values
#
voxpelli
But that would also mean that one could probably do the entire IndieAuthProtocol thing without showing a single web view?
#
ben_thatmustbeme
voxpelli: huh, don't know about ios dev
#
aaronpk
twitter auth isn't required for indieauth protocol
#
ben_thatmustbeme
my app is apache cordova, so its written in html/css/js and just compiled to android (and ios if anyone cares to do that)
#
ben_thatmustbeme
Jeens, ask benwerd on that one though
#
ben_thatmustbeme
s/Jeens/Jeena
#
Loqi
ben_thatmustbeme meant to say: Jeena, ask benwerd on that one though
#
loqi.me
created /Ben (+51) "prompted by Zegnat and dfn added by ben_thatmustbeme"
(view diff)
#
voxpelli
aaronpk: right, so one would always have to use the web unless one could detect that a specific web endpoint had a matching app :P
#
aaronpk
the first step is to look at the person's home page and find their authorization endpoint
#
aaronpk
so you'd just send them there
#
voxpelli
thinking about if he can convince one of the iOS devs he knows to code this
#
aaronpk
it's that authorization endpoint that may have support for twitter login
gavinc joined the channel
#
voxpelli
doesn't look that hard to add such a provider
#
aaronpk
i guess what i'm saying is the app won't know whether the auth server supports twitter or whatever
tilgovi and cm100 joined the channel
#
voxpelli
aaronpk: absolutely, I'm thinking beyond that now – if one could just make something that handles all of the rest of indieauth stuff so that one can get a simple passport.s / omniauth-like flow where in the end one just gets a token and doesn't have to care about how one arrived there
#
aaronpk
like as an ios library?
#
voxpelli
such libraries will probably be needed to get regular apps on board with micropub as many developers thinks regular OAuth dancing is hard enough
#
aaronpk
gotcha. yeah that'd be great to have as a library
#
aaronpk
also the new safari view controller thing helps a lot, since it opens a real safari window inside the app instaed of popping you out of the app
#
voxpelli
finally OAuth on iOS (and Android who will get the same) can be considered secure :P
#
Jeena
yeah I hated it that every app just showed my their own webview which they controlled instead of using safar/firefox/chrome
tantek, snarfed1, benwerd, wolftune, ttepasse_, j12t, cm100, parzzix and friedcell joined the channel
#
kevinmarks.com
edited /Webmention (+205) "/* Why is the target URL a required parameter */"
(view diff)
#
gregorlove.com
edited /Webmention (+0) "/* Why is the target URL a required parameter */ typo"
(view diff)
KevinMarks joined the channel
#
paultibbetts.uk
edited /WordPress/Examples (-71) "/* Other independents using it on their primary site */ Paul Tibbetts new domain"
(view diff)
snarfed, wolftune, benwerd and soypirate joined the channel
#
grant.codes
edited /2015/Edinburgh/Guest_List (+281) "/* Creators */ Add Grant Richmond"
(view diff)
#
grant.codes
edited /2015/Edinburgh/Guest_List (+2) "Update attendee numbers"
(view diff)
tantek and KartikPrabhu joined the channel
#
tantek
fellow HWC organizers aaronpk, kylewm, benwerd, snarfed, GWG in looking at the current page, https://indiewebcamp.com/events/2015-07-15-homebrew-website-club I feel like URLs & RSVP could be combined into one section, RSVP - what do you think?
#
benwerd
+1 to that
#
Loqi
benwerd: tantek left you a message 1 day, 3 hours ago: what do you think about adding person-tagging support to photo posts in Known? I think that plus supporting salmentions would upgrade Known to be able to play player A's role in https://indiewebcamp.com/SWAT0#implementation_requirements http://indiewebcamp.com/irc/2015-07-14/line/1436900658124
#
Loqi
benwerd: tantek left you a message 4 hours, 27 minutes ago: note comments markup updated slightly with u-comment instead of p-comment, see https://indiewebcamp.com/comments-presentation#How_to_markup for update and reasons! http://indiewebcamp.com/irc/2015-07-15/line/1436982278421
#
tantek
(was about to post a reminder for all HWC meetsup tonight, thinking to link to the "RSVP" and realized it misses out on the chance to Indie or FB RSVP
#
aaronpk
agreed
#
tantek
that diagram is crazysauce
#
tantek
every relationship is just labeled "annotation"
#
tantek
not a useful use of that term
#
tantek
if you can s/annotation/hyperlink and not lose any meaning, then your use of "annotation" had no meaning to begin with
#
benwerd
Least cute cat-based Internet example ever :/
#
tantek
benwerd - time to add cats to fatberg?
#
KartikPrabhu
should get to getting likes/reposts working with marginalia :)
#
tantek
KartikPrabhu++ yes!
#
KartikPrabhu
after hfeed2atom is in some shape
#
Loqi
KartikPrabhu has 101 karma
#
tantek
thanks benwerd & aaronpk, here goes nothing
#
benwerd
tantek: catberg
#
benwerd
(they deserve their own monolithic standard)
#
tantek
benwerd, sounds like catbert
#
KartikPrabhu
wow in that diagram everything is actually just a mention even without hyperlinks :|
#
benwerd
it could be an annotation on catbert
#
tantek
likes making flat lists (or "collections" heh) of different kinds of things, it really annoys the data modelers.
#
KartikPrabhu
Feedly jus told me it is making a "collection" UI
#
KartikPrabhu
opt-in to try it
#
csarven
tantek s/mention/hyperlink ;)
#
tantek
csarven, pretty much
#
KartikPrabhu
mention is a hyperlink
#
tantek
KartikPrabhu: are all hyperlinks mentions?
#
tantek
is mention merely a synonym?
#
KartikPrabhu
tantek: sure why not. that is how I treat them if I receive a webmention. if just hyperlink no additional markup then yes
#
tantek
csarven, have you been following all the W3C Annotations stuff?
#
tantek
KartikPrabhu: agreed, I think that's reasonable
#
tantek
could easily see "Other links to this page" as alternative copy to "Other mentions"
#
tantek
I think "mention" is just shorthand for that
#
KartikPrabhu
tantek: yes
#
KartikPrabhu
I was thinking of having a "related articles" list if the mention is a long article... but not gotten to that yet
#
csarven
Kilicdaroglu, what a last name..
#
csarven
is looking at doe's PNG
#
csarven
tantek re: w3c annotations, only the data model.
#
tantek
csarven - how about the use-cases?
#
csarven
the protocol "feels" a bit overkill.. I don' tknow
#
tantek.com
edited /events/2015-07-15-homebrew-website-club () "(-846) collapse URLs section with RSVPs for simplicity"
(view diff)
#
csarven
tantek No, I haven't looked at the UCs yet. What I do know is along the lines of what you said, it casts a wide net on what can be an annotation.
#
tantek
csarven, reminds me of "cards" and how everything is (or can be?) a card
#
aaronpk
looks nice
#
csarven
Having said that, this http://www.w3.org/annotation/diagrams/annotation-architecture.svg which you've probably seen is a brief walk-through.
#
benwerd
tantek: looks way better to me. I wonder if there's scope to collapse the URLs into links, so that it's just the RSVPs and a compressed link to the indie event / facebook rather than bullets. But this is excellent
#
tantek
benwerd thank you as yes absolutely feel free to further iterate and improve it!
#
tantek
s/as yes/and yes
#
Loqi
tantek meant to say: benwerd thank you and yes absolutely feel free to further iterate and improve it!
#
csarven
opens up his notebook -- yes the paper/pen thing.
#
@t
Reminder: HWC #IndieWeb meetup SF+Portland TONIGHT! 17:30 writing hour 18:30 chat & hack night https://indiewebcamp.com/events/2015-07-15-homebrew-website-club#RSVP (ttk.me t4cB1)
(twitter.com/_/status/621446934604214273)
#
tantek
paper++
#
tantek
pen++
#
Loqi
paper has 2 karma
#
Loqi
pen has 1 karma
#
csarven
I classify annotations/comments/mark/highlight/*note (footnote, marginal, endnotes..) to have high overlap.
#
csarven
citations/blockquote/q/anchors are somewhat related to annotations but are sufficiently different.
#
tantek
awaits csarven's Venn diagram ;)
#
KartikPrabhu
csarven: that is fine as a concept until you try to implement a UI for them
jciv joined the channel
#
csarven
KartikPrabhu That is exactly what I'm doing! :)
#
csarven
Will find out in due time.. like I said, this is something I sketched in my notebook while wondering around..
#
csarven
Thanks for the pointer.
#
csarven
I think in my case, it will mostly be for small snippets.
#
csarven
(if not all cases come to think of it)
#
tantek
small snippets are a good place to start. I hear you can build a whole multimillion user platform on small snippets. ;)
#
csarven
Standard scholarly articles.
#
csarven
hah
#
csarven
considers standardizing on 10 chars max.
#
tantek
^^^ self-violation of consideration. ;)
#
csarven
Took me a second to get that!
#
tantek
I am kinda curious what the say 90th percentile of IRC statement length is.
#
csarven
That'd be cool. Cluster around type of channels.
#
csarven
Tech, social, news..
#
aaronpk
has lots of IRC logs to grep
#
csarven
Correlate with average age and language
#
csarven
tantek Not from OA but from DPUB: http://www.w3.org/TR/dpub-annotation-uc/
#
csarven
"in coordination with the oA"
snarfed joined the channel
#
csarven
Can't go through that now. Insufficient brain cells for today.
lukebrooker joined the channel
#
csarven
I'm constantly logged out of IWC.com, is it just me?
#
aaronpk
there's a preference in your user page
#
csarven
Thanks
#
tantek
goes back to benwerd's design feedback.
#
tantek.com
edited /events/2015-07-15-homebrew-website-club (+35) "/* RSVP */ try some link compression with SF as the test case (we're good at that)"
(view diff)
#
tantek
benwerd, here's a first attempt at following your suggestion, what do you think? https://indiewebcamp.com/events/2015-07-15-homebrew-website-club#San_Francisco
#
tantek
(left Portland the way it was so you could compare)
#
benwerd
+1 from here - looks much easier to read for me
#
tantek
great! completing design transition...
#
tantek.com
edited /events/2015-07-15-homebrew-website-club (+41) "/* RSVP */ Portland link compression"
(view diff)
#
tantek
... design transition completed.
#
tantek
gRegorLove: heads-up, major structural changes in the sections / headings of today's HWC as compared to before - may want to reset your clone base to it.
#
tantek
have also dropped all commented out locations, and we can add them back in as needed.
#
kylewm
tantek: do you want (or want me) to update 07-29 too? https://indiewebcamp.com/events/2015-07-29-homebrew-website-club
#
tantek
kylewm: go for it - thank you!
#
gRegorLove
Sure, I'll use the latest version if I set up more HWC pages. If others want to, though, that's cool. I'm a bit out of the HWC loop :)
#
kylewm.com
edited /events/2015-07-29-homebrew-website-club (+174) "/* RSVP */ update to NWO"
(view diff)
#
kylewm.com
edited /events/2015-07-29-homebrew-website-club () "(-538) /* URLs */ collapse URLs section into RSVPs"
(view diff)
#
tantek
gRegorLove: you know all your help with the wiki is appreciated. Saw you got a bunch of the Etherpads archived as well - and thank you for updating the /2015/Schedule to point to all the sessions / archives!
#
tantek
gRegorLove++
#
Loqi
gRegorLove has 18 karma
#
gRegorLove
Appeals to some administrative thing in me that likes to organize stuff, heh
#
tantek
you're definitely among friends in that respect too :)
scor joined the channel
#
tantek
speaking of dates (ref: earlier conversation about talks being given, video *recorded* date, video *published* date), I'm now having a challenge with my music re: dates
#
tantek
as another data point on this created vs. published problem
#
gRegorLove
Hmm, not the best sign if the Western Washing University ACM subdomain doesn't resolve (linked from the CS page). I was thinking about reaching out to see if there's any interest there for a Bellingham HWC
#
tantek
iTunes comes with a "90's music" 'smart' playlist, so of course I created a similar "80's music" 'smart' playlist and that's where my trouble started.
#
tantek
I have a CD Single of New Order's Bizarre Love Triangle that I've ripped, and it says "year:1994"
#
gRegorLove
Sure it didn't start when you downloaded iTunes? ;)
#
tantek
of course that's wrong, the song (and that specific version of it) was released in 1987
#
tantek
gRegorLove: I don't think I ever downloaded iTunes - it was always bundled with the OS
#
tantek
er, s/1987/1986
#
tantek
Release date (1994) vs. Copyright (1986).
#
tantek
so now, as gRegorLove put it, that "thing in me that likes to organize stuff", is having difficulty dealing with this incongruity, especially when there is only one "Year" field in iTunes, and *obviously* Bizarre Love Triangle belongs in the "80's music" smart playlist, not in the "90's music" smart playlist.
#
gRegorLove
I used and contributed to musicbrainz.org for meta data when I was ripping music more regularly
#
tantek
to tie this into something we talked about at IWC PDX, I can see publishing parts of my music collection this way (everything but the actual audio), and thus how should I do so?
#
gRegorLove
They have some free tagging software too that thumbprints mp3s and can be used to update the mp3 metadata
#
tantek
especially if I had scripts to convert to/from iTunes (I think Chloe was working on that part)
#
tantek
(wow way too much apophenia here, must have been that TCHO chocolate I just had)
#
gRegorLove
Yeah, I was really impressed by Chloe's page with her listening history (I think that's what it was?)
#
tantek
gRegorLove: yes, listening history, and extensive personal tagging system she had for I think seasons, moods, etc.
#
gRegorLove
Think iTunes library stuff is in XML, so I'll leave that to someone else. :)
snarfed joined the channel
#
tantek
gRegorLove: it's a good question whether to scrape / sneak the XML underneath iTunes or to attempt to use the AppleScript "API" to access it.
#
tantek
would love to (eventually, not an urgent itch) to come up with a way to post an indie mix tape (and yes I know all the prior XSPF art in this area).
#
tantek
heck, even post indie web versions of the mix tapes I actually made in the 1980s.
#
tantek
with full scan of the custom labels I printed of course
#
tantek
is anyone posting indie web mix tapes?
#
tantek
ok, back to the original Released vs Copyright date
#
tantek
I think I have an answer
#
aaronpk
released sounds like "published" to me
#
tantek
in this case it's more like "republished"
#
gRegorLove
I shared a zip file of a mix I made years ago, though the zip wasn't self-hosted. One of those free upload sites.
#
tantek
as it was originally published in 1986 (though on 12" - which I also have)
#
tantek
gRegorLove++ awesome
#
gRegorLove
(still rather proud of that mix, and I'm overdue to make another)
#
tantek
yes that post of yours is exactly what I'm talking about - all the info about the tracks
#
gRegorLove
obviously, do so only at your own copyright risk, heh
#
tantek
but without direct hosting of the music
#
tantek
what is a jam?
#
tantek
hmm - apparently I didn't document that
#
tantek
is a play list just a specific kind of /collection?
#
tantek
mix tape = play list etc.
#
tantek
although a collection implies everything in it has its own permalink, whereas a play list might not (e.g. gRegorLove's example)
#
tantek
popping the stack back to I think I have an answer
#
tantek
back when I used a HyperCard stack to catalog all my music, I would also only keep one year - but I specifically labeled that year "Copyright", because, that's what was actually *printed* on the CD (or other medium).
#
tantek
capturing the visible data as it were
#
tantek
thus I'm going to treat the iTunes "Year" field as Copyright as well
#
tantek
and move the "Released" date into the Notes field.