2017-07-27 UTC
leg1, jjuran, tantek, [kevinmarks], Defenestrate, j12t, [eddie] and davidmead joined the channel
j12t and davidmead joined the channel
jjuran, tantek, eli_oat1, gRegorLove, j12t, [miklb] and cweiske joined the channel
KevinMarks, jjuran, [acegiak_net] and [chrisaldrich] joined the channel
jeremycherfas joined the channel
AngeloGladding, Defenestrate, barpthewire, KevinMarks and [kevinmarks] joined the channel
KevinMarks joined the channel
KevinMarks joined the channel
tantek joined the channel
KevinMarks_, eli_oat1, jeremycherfas, KevinMarks, [kevinmarks], [eddie] and [dgold] joined the channel
# 12:39 sknebel maybe we should consider scope editing for selfauth?
# 12:46 Zegnat We can easily offer the checkboxes, like aaronpk does, yes
KevinMarks joined the channel
# 12:50 sknebel for the issue mentioned (post vs create) the ability to add would be good as well
# 12:50 sebsel I support an extra scope myself, 'media', wether or not people may use my media endpoint.
# 12:50 sebsel So there are two strategies there: show the user all the requested scopes and ask yes/no
# 12:51 sebsel Or let the user define a list of scopes they offer, and auto-check matching boxes
# 12:51 sebsel I do the second one now, and I think aaronpk does too
# 12:53 sebsel Micropublish.net requests 'undelete', but I don't have that, so I don't give it.
# 12:54 Zegnat I am not sure how I feel about needing people to define a list of supported scopes on setup though.
# 12:54 sebsel It can be a default list, somewhere, that is editable when the user really wants.
# 12:54 Zegnat Giving the option to yes/no and add extra values may be enough for the selfauth usecases
# 12:54 Zegnat There are no default values defined by IndieAuth though
# 12:57 sebsel Yeah, I'm not saying that's the best thing to do for SelfAuth, but those are two strategies to think about.
# 12:57 Zegnat Default list needs to come from somewhere. The Micropub spec hardlinks to https://indieweb.org/scope for “a list of all currently used values”. That’s why I would like to shy away from that. (My opinion can of course be changed, will see where the discussion goes.)
# 13:00 Loqi [sknebel] #26 Allow changing the granted scopes
# 13:02 voxpelli Feels like the /scope needs to be extended with more examples especially – seems like a few here are using more scopes than documented there? Only me + aaronpk in there now
# 13:03 Zegnat voxpelli, I agree, especially if the W3 Micropub spec links to it for a list. I have nothing to add there though.
# 13:14 sebsel but in the end, it's just the question of 'can I do this'?
# 13:16 sebsel The app just hints what it thinks it might need.
# 13:16 sebsel I can also use scopes in Dutch, like 'bewerk' for 'update'. I just need to make sure to give the app the scope 'bewerk' if it asks for 'update'.
# 13:17 sebsel It is my site that checks which operation is allowed by which app at what point.
singpolyma joined the channel
# 13:39 voxpelli it's fairly good to try do standardize on a fairly limited set though so apps can clearly state what scopes they need for what operations and that way apps can be given as strict access as possible
# 13:39 voxpelli else it can easily be that all apps are given all access and the point of scopes pretty much is lost
# 13:40 voxpelli (not a problem with current indieweb generation, but with future generations and as things scale it can be)
# 13:41 sebsel also: if the app knows which scopes it has, it can already anticipate a 403 if it did not receive a, for example, update scope.
# 13:42 Zegnat Yes, that’s why apps get a list of scopes returned to them when they verify the indieauth code. So they know exactly what scopes they were granted and follow through accordingly
# 13:52 voxpelli yeah, should be possible to eg. deactivate update features and instead show an "give update access to do X"
[barryf] joined the channel
# 14:01 jeremycherfas Anyone got any good examples of styling webmentions from voxpelli’s Heroku app? I’d like to have a starting point with which to tinker.
KartikPrabhu, KevinMarks, KevinMarks_ and [kevinmarks] joined the channel
KevinMarks joined the channel
# 14:26 jeremycherfas Voxpelli The classes on your webmentions differ from the ones I got in the examples. Is that because you are using bleeding edge?
j12t joined the channel
# 14:27 voxpelli jeremycherfas: Was a while since I worked on that specific code, it may be that I cleaned up the class names in bleeding edge (better not to use bleeding edge right now though, it's a work in progress to support Salmentions)
KevinMarks_ and KevinMarks joined the channel
# 14:39 dgold sorry to ask such a basic question but, checking for either scope 'create' or scope 'post'
# 14:40 dgold if(!strinstr($scope, 'create') || strinstr($scope, 'post) { 403 }
KartikPrabhu1 and KevinMarks joined the channel
# 14:44 Zegnat I don’t know what your $scope variable looks like
# 14:45 dgold scope is from indie-auth response - I'm trying to do a check so that if the scope has neither 'create' nor 'post' then the endpoint serves a 403
# 14:45 Zegnat if (strpos($scope, 'create') === false && strpos($scope, 'post') === false) { /* the string 'create' and the string 'post' do not exist in $scope */ }
# 14:46 Zegnat Aah, IndieAuth scope value, then there is a slightly better one, give me a second
# 14:50 Loqi zegnat has 23 karma in this channel (121 overall)
# 14:50 Zegnat strpos could have false positives, e.g. a scope "imposter" would have matched as including "post". This way you get around that.
# 14:50 Zegnat If for whatever reason I made a typo in that code or it ends up not working you know where to find me ;)
# 14:50 dgold the !notation and either/ors make my head hurt
# 14:57 Zegnat in_array() will give false when the string isn’t there, and in this case you want to test for the false result so you need the ! (or `=== false` comparison). And then you want to do that twice and not have it exit after the first test (thus &&)
[miklb], j12t, KevinMarks_ and eli_oat joined the channel
# 15:51 jeremycherfas Voxpelli: Is there a way to test whether the script that pulls in webmentions contains anything, so I can modify the display conditionally?
KartikPrabhu, [miklb] and KevinMarks joined the channel
# 16:28 voxpelli jeremycherfas: don't think so with current version, add an issue and I'll try to make it so eventually :)
# 17:02 dgold is there a media-endpoint exemplar available? (prefereably in php)
KevinMarks joined the channel
[miklb] and gRegorLove joined the channel
j12t and KartikPrabhu joined the channel
# 17:30 sebsel Ah, nvm me. I have the right code in my code, just not in my head.
tantek joined the channel
[kevinmarks] joined the channel
tantek joined the channel
j12t joined the channel
cweiske joined the channel
# 18:45 dgold aaronpk: sorry to ask such a question but; if i'm indieauth'd onto one site - example.com, and micropubbing to that site, and I need to upload to a media endpoint - media.example.com - is that a separate 'login'?
# 18:47 schmarty dgold: the media endpoint should accept the same token as your main micropub endpoint
# 18:49 dgold that's the bit I'm not understanding, schmarty
# 18:49 dgold looking over the endpoint aaronpk posted to github - I can't see where that logic resides?
# 18:51 dgold but media.ascraeus.org is a different 'location'
# 18:52 schmarty so the implementation that aaronpk linked requires an authorization token in an HTTP Authorization header
# 18:53 schmarty and it checks it against tokens.indieauth.com to make sure it is still valid and that it has a scope
# 18:53 sknebel but it never checks for the identity the token is for? (aaronpk)
# 18:53 schmarty this implementation is not doing any verification about the "me" value that i can see
# 18:54 dgold yes - I see where it looks for the token, line 15+, and I see where it starts checking it, line 37
# 18:54 schmarty dgold: the micropub spec says the media endpoint should accept the same tokens that a "regular" micropub server endpoint should accept
# 18:55 dgold but I don't see that its checking the token is valid for the originating location - the one without [media] in the W3C spec
# 18:55 schmarty so in your case it should accept tokens for "ascraeus.org"
# 18:55 schmarty dgold: this implementation seems not to check that value at all
# 18:57 aaronpk so, I modified the gist a little from what my own implementation does
# 18:57 dgold there's loads of catpics being uploaded to your media endpoint
# 18:57 aaronpk I have my own token endpoint so it will only return 200 from the token check for my own tokens
# 18:58 aaronpk so yeah if you use a shared token endpoint then you'd need an additional check there looking at the response from the token endpoint
# 18:59 dgold and that would be checking a 'me' value for the originating mpub location?
# 19:01 dgold phew - that explains that. I'd been puzzling at this for far too ling
# 19:06 Loqi aaronpk has 68 karma in this channel (1390 overall)
KevinMarks, j12t and tantek joined the channel
j12t and j12t_ joined the channel
# 20:37 Loqi aaronpk has 69 karma in this channel (1391 overall)
KevinMarks, tantek and j12t joined the channel
# 21:16 gRegorLove Ooh, just caught up with indieauth.com getting renamed to indielogin.com
# 21:17 aaronpk i'm hoping to be able to focus on that project in august
# 21:36 Zegnat Hope you’ll think of us when you need some eyes on it, aaronpk ;)
# 21:37 sknebel Zegnat: right, we should look at moving your draft over again, or at least large parts of it. or wait for the spec, cross-check everything and then?
KevinMarks joined the channel
# 21:39 Zegnat That note isn’t in my draft page, because hopefully the draft page wouldn’t need such ntoes
# 21:42 gRegorLove Draft page looks good. I think "Developers can use any way they want of finding these." probably isn't necessary.
# 21:43 gRegorLove Those can also be in HTTP headers, right? So maybe list those variants there.
# 21:43 Loqi zegnat has 24 karma in this channel (122 overall)
# 21:47 Zegnat True, that developer specific sentence was probably unnecessary, gRegorLove. Feel free to edit if you have any ideas. It’s not just “my” draft.
# 21:49 aaronpk we should probably do a survey of how many people advertise in Link headers, and not include that in the spec if nobody's using it
# 21:50 Loqi zegnat has 25 karma in this channel (123 overall)
# 21:51 sebsel It's actually easier in CMS plugins, because you just need to call header(), instead of asking the user 'add this to' or finding the right hook to do so.
# 21:51 Zegnat I think it is great to be able to tell to people to copy an HTML link element for set-up. But for development headers are so nice and easy. Get them through a HEAD request, add them to server config and instantly have them available site-wide.
# 21:52 sknebel well, the auth_endpoint you odn't need sitewide, but still, no editing templates
# 21:52 Zegnat And unlike with webmention endpoints where you might want to really scope it to separate pages, I think most people will just use a single authorization-endpoint for everything so it could be added to server configs without any penalties
# 21:54 sebsel I was going to say: for private webmention you need site-wide, but that's the token endpoint.
# 21:55 Zegnat IndieAuth defines working together with the token-endpoint. I don’t think token-endpoint is IndieAuth specific though.
KartikPrabhu and tantek joined the channel
# 22:31 sebsel I have a basic webmention plugin for Kirby again :)
# 22:31 sebsel Not live on my site, still a lot to tweak, but I like what it does now
# 22:32 sebsel It uses a lot of aaronpk's code, including XRay, php-mention-client and even QuartzDB, but oh.
# 22:33 sebsel I store all webmentions in QuartzDB, which is just a file with a JSON-with-date per line
# 22:33 sebsel And then on every post I store a dictionary of URL to ID in the database (date and line number)
# 22:34 sebsel but a new mention overwrites the pointer, thus updates it.
raucao, Ruxton and [kevinmarks] joined the channel
sebsel and AngeloGladding joined the channel
# 23:59 Loqi [gRegorLove] mf2-to-iCalendar: Convert microformats h-event to iCalendar