2017-08-15 UTC
# 04:56 xmpp-social [ajordan] Ayyyyyyyy # 04:56 xmpp-social [ajordan] That's awesome timbl_ joined the channel
# 06:24 xmpp-social [ajordan] Hey the wiki says we have a meeting tomorrow but I have a distinct memory of Evan saying "last chance to uncancel the meeting on the 15th" # 06:24 xmpp-social [ajordan] ???????? xmpp-social and dwhly joined the channel
# 06:58 rhiaro evan gave us last chance to cancel the 8th and the 22nd I thought # 06:59 rhiaro according to that, tantek is chairing and evan will be present # 06:59 xmpp-social [ajordan] Ah that must be it. Thanks rhiaro! # 06:59 xmpp-social [ajordan] Been reading SWP btw :-) # 13:34 Gargron the one other thing i want to raise but might be too late to raise: WebSub subscriptions have an expiration date, so they get renewed periodically, so you eventually get rid of dead endpoints. but in AP, follow is forever-until-unfollow. it'll require some sort of keeping count of failed deliveries and marking accounts as dead # 13:40 saranix It is important to remember that, and this goes double for indie social sites, there is a huge range between alive and dead. Therefore an exponential backoff or dead-cliff is insufficient. # 14:29 cwebber2 Gargron: it's not too late to raise it... file an issue. # 14:30 cwebber2 esp because we had a resolution that we wouldn't add new things to AP that weren't adjustments to the protocol # 14:30 Gargron i dont have an obvious solution. like... what do you do? make Follow activities expire, and re-send them periodically? that seems kinda weird in AP context # 14:30 cwebber2 Gargron: I agree it does seem weird. I would think that occasionally polling for liveness is a good idea. # 14:31 cwebber2 and if something appears non-alive, give some range of time you re-check until you mark it as dead # 14:42 cwebber2 Gargron: btw there will be a socialcg call tomorrow if you want to make it; should I put this on the agneda? # 15:55 Gargron cwebber2: what can I do if my users Announce a note that originally came from OStatus, and has a tag:example.com;etc URI? # 15:56 cwebber2 hm, in a meeting but I need to think about it. could you file an issue in the AP repo and I'll reply to it? tantek joined the channel
# 16:25 wilkie follow liveness is complicated. something not being alive during a renewal window doesn't necessarily mean you want to get rid of it. # 16:26 wilkie you can still follow someone or have a record of a relationship and have them not really exist and thus not get their feed... leaving the semantics of pulling the feed or not to the federation protocol, which is better suited to deal with this issue # 16:27 tantek I tend to agree with wilkie. there's a difference between the user semantic of, I've established a follow relationship, and the plumbing guts of "subscription periodically expires, renew" # 16:27 cwebber2 wilkie: yeah, my sense is that you want to do some sort of series of polls over a period of time, and if they fail to appear during all of them (maybe over the course of 6 months or a year) they get removed # 16:27 tantek to design the user model based on the plumbing conceptual model is also an error # 16:28 wilkie yeah, I tend to like the separation of concerns of the graph/knowledge protocol and the federation/pubsub protocol # 16:29 wilkie cwebber2: I think there is a good rationale separating "well, I guess I don't poll this feed anymore" and "well, I guess I don't follow this person anymore" governing what the "they" is in your sentence that gets removed # 16:30 cwebber2 is a bit confused by that sentence's distinguishment of "they" # 16:30 puckipedia ok, finally got a bit over my illness, let's try to test AP in Mastodon # 16:31 puckipedia wellll technically I still have it; I'm just not bascially limited to doing nothing with it # 16:32 puckipedia I should build a good logging system into Kroeg hmmm # 16:32 cwebber2 the blocking issue is fixed; I noticed that before some things are passing but when I tried yesterday after I put it up again I was hitting a bunch of 500 errors # 16:32 puckipedia and then some way to link logs to requests; and then show a list of requests and their corresponding logs # 16:32 rhiaro I would have no friends left on facebook if it deleted them because I haven't posted in a year. The way subscription works in AP is that new posts are pushed though right? So we there's nothing to back off from? It's okay to have someone appear in your friends list but never receive any updates from them. # 16:33 cwebber2 specifically, use the AP client-to-server server tests, the other ones aren't implemented yet (but I'm workin' on it) # 16:33 puckipedia rhiaro: I think the issue is more "what if the federating server consistently rejects your messages or is offline for a long time?" # 16:33 cwebber2 yeah it's up! I managed to get past the plumbing issue through terrible hacks :) # 16:33 puckipedia e.g. the server hosting a bunch of followers died # 16:34 rhiaro yeah I don't think they should disappear off the followers list just because they aren't getting things. Internal mechanism to decide when to stop bothering to send, perhaps. Ie. implementation detail, that could come with a note of suggestion (ie. non-normative) perhaps # 16:34 tantek and in that case, how does one reconnect to those followers on another server? # 16:35 puckipedia this would be a server->client notifying system anyways # 16:38 jaywink I think the way diaspora does the "a node died, we should probably not do timeouting requests forever" is to keep a last success timestamp. If the node doesn't have any successes for X days -> it is not posted to any more. But no actual data is changed, it's just a server implementation thing, not protocol. This is also how my relay system works. # 16:39 jaywink it would seem odd removing followers because the node has gone down, because then you make this permanent. A failed timestamp could be cleared for example when receiving a post from the node. # 16:39 rhiaro contemplates if she has time to make a pizza before the meeting # 16:39 jaywink I know nodes that have been down for a very long time. Someone just got bored for a while but then brought everything up again later # 16:39 wilkie a machine shouldn't manipulate your friendships lol # 16:39 puckipedia so; Kroeg has a small fallback mechanism built in which is relevant to this # 16:39 rhiaro and by make a pizza I mean starting from flour and water # 16:40 cwebber2 rhiaro: in 20 minutes, that would be fast, but I guess feasible! # 16:40 puckipedia it tries to cache objects that it doesn't own (remote servers etc) for up to 1 week # 16:40 puckipedia afterwards, if you try to get it, it will see that it 'expired', try to get it from the server; but fallback if that isn't possible # 16:41 puckipedia Gargron: hm, will Mastodon use ActivityPub when talking internally? # 16:41 Gargron puckipedia: yes # 16:41 puckipedia as in, is that in current master? # 16:42 puckipedia oh it is lol # 16:42 rhiaro In other news, for those who like PHP and RDF (haha that's just me), I finally got most of the way to extending EasyRdf's JSON-LD serialiser to serialise to conformat AS2 if you ask it for activity+json # 16:43 puckipedia Gargron: ... does that mean I could rip out all the OStatus parts from Mastodon and still have it work? lol # 16:43 Gargron puckipedia: yes, in the future # 16:43 puckipedia (also can it migrate existing Mastodon accounts to use ActivityPub?) # 16:44 puckipedia (like, when it re-retrieves the user info, does it try to pick up the ActivityPub version?) # 16:47 puckipedia ooh it seems to do so :D # 16:49 puckipedia yarn and bundler aren't really fast in WSL # 16:57 xmpp-social [ajordan] I can't believe rhiaro is making a pizza from scratch and here I am, very jetlagged, barely awake enough to pour myself granola # 16:57 xmpp-social [ajordan] ;) # 16:58 tantek oops I just committed my WebSub implementation report directly without creating a pull request # 16:59 rhiaro did not manage to roll out the dough yet, but it's made and the toppings are chopped. So close. # 17:01 aaronpk i don't know why i haven't been able to memorize this new telcon code. i had the old one memorized # 17:01 tantek speaking of which also updated my Hangouts app so let's see - yes it works # 17:01 rhiaro declines from scribing duty on account of pizza duty # 17:01 ajordan aaronpk: I just have a "SocialWG telecon line" which includes the meeting code RRSAgent joined the channel
Zakim joined the channel
# 17:02 aaronpk ajordan wait what, including pressing the numbers after it picks up? # 17:03 cwebber2 it wouldn't be the socialwg if I didn't dial in 2-3 minutes late # 17:03 rhiaro aaronpk: I have an email in my drafts so I just search 'webex social' whenever I need to call # 17:04 Zakim Present: ben_thatmustbeme, rhiaro, tantek, aaronpk, ajordan, sandro, tsyesika, cwebber # 17:04 Zakim sees on irc: RRSAgent, tantek, dwhly, xmpp-social, timbl_, nightpool, bwn, trackbot, cwebber2, jungkees, wilkie, dlongley, MMN-o, rhiaro, jet, ben_thatmustbeme, csarven, jaywink, # 17:04 Zakim ... bigbluehat, sknebel, tsyesika, astronouth7303, Loqi, Gargron, puckipedia, raucao, tcit, ajordan, mattl, DenSchub, saranix, jankusanagi_, aaronpk, KjetilK_, bitbear, albino, eprodrom joined the channel
# 17:05 ben_thatmustbeme most important part is to review for resolutions that you might not have been there for and want to raise issue with i believe # 17:07 ajordan tantek: I'll declare the minutes resolved; I'll have to read about the yak shaving later # 17:07 eprodrom we hereby resolve to propose to resolve... # 17:08 ajordan cwebber2: good news! test suite is up at last and I can demo it live # 17:08 ajordan ... last two weeks I was complaining about yak shaving, I had to fix stuff in my language # 17:09 ajordan ... you can visit this, put in this username, and give this token # 17:10 eprodrom Wonderful job cwebber2! # 17:10 eprodrom cwebber2++ imts # 17:10 ajordan ... c2s and ??? has to be done but I have the general framework and am moving forward # 17:10 ajordan ... even better news, apparently Mastodon has AP mostly implemented # 17:10 ajordan ... they are working on testing things out now and that's also good news # 17:11 ajordan ... things are moving along, 2/3 good news things but I'll pause for Evan # 17:11 ajordan eprodrom: cwebber2 I want to make sure, Mastodon is just doing the s2s protocol right? # 17:11 ajordan eprodrom: so we can't test what Mastodon is doing # 17:11 ajordan cwebber2: yes, I've tested puck's stuff and have seen things succeeding and failing for logical reasons # 17:12 ajordan cwebber2: anyway I'll work on getting the s2s tests, I'll try to prioritize those over c2s especially because of Mastodon # 17:12 ajordan ... the tests, in the UI there was the c2s test and the s2s test and then a separate federation test? # 17:12 ajordan cwebber2: these are the different spec profiles in the AP spec # 17:12 ben_thatmustbeme wonders if Mastodon going with support for AP and ostatus, would it be easier to add other non-internal federation protocols webmention / mf2 for example (cc: Gargron) # 17:13 ajordan ... c2s client -> clientside, c2s server -> API, s2s API # 17:13 ajordan ... currently working on c2s because that's the easiest to do # 17:13 ajordan ... I wanted to make sure I did the one that gave me confidence things were working # 17:13 ajordan tantek: so you have the c2s clientside ones done? # 17:14 ajordan ... there's some server stuff worked in there because clients can do some addressing and they don't know if it works, so that's marked as unknown # 17:14 ajordan tantek: re: other variants, do you have an estimate? next week or two? # 17:15 ajordan cwebber2: trying to go as fast as I can, it's difficult because I'm best man in a wedding # 17:15 ajordan ... think we'll have s2s tests up within the next month and client tests should be really fast to do after that # 17:15 ajordan cwebber2: I wanted to publish a new... there've been some non-normative change # 17:16 ajordan ... and we've talked about some normative changes # 17:16 ajordan ... thought it would be good to publish a CR with a lot of non-normative typo fixes before that # 17:16 ajordan ... some stuff added to the security considerations but those are non-normative # 17:17 ajordan tantek: that sounds like a good idea, it's been four months so that's another reason # 17:17 ajordan ... these changes are all non-normative and won't affect implementations? # 17:17 ajordan cwebber2: yes, security considerations might affect implementations but they're non-normative # 17:18 eprodrom someone using an IBM Model M keyboard? # 17:19 ajordan eprodrom: I'm surprised you haven't noticed it by now :P # 17:19 tantek also: s/E.1 Changes from 13 April 2017 to present/E.1 Changes from 9 May 2017 to present # 17:20 ajordan I'm not seeing the Apr. 13 change showing up on GitHub Pages? # 17:21 ajordan tantek: I think because it's a CR we still need staff to turn the cranks for us # 17:21 ajordan ... I think we noted that in the agenda, presumably rhiaro... yeah August 22nd publications resume # 17:21 ajordan ... because it's a CR we have to send an email for approval to update # 17:22 ajordan tantek: I believe rhiaro said she was just going to take care of it # 17:22 ajordan ... so we'll expect that published Tuesday morning of next week # 17:22 ajordan ... we don't have a meeting but at least something happens for us! # 17:22 ajordan ... any issues you wanted to discuss for normative changes? # 17:23 ajordan cwebber2: no, I want to make changes that we discussed last meeting and then discuss those # 17:23 ajordan tantek: I notice informally you've been steering people towards extensions, that's good for the CG # 17:23 Zakim tantek, you need to end that query with '?' # 17:23 Zakim Present: ben_thatmustbeme, rhiaro, tantek, aaronpk, ajordan, sandro, tsyesika, cwebber, eprodrom # 17:23 Zakim sees on irc: eprodrom, Zakim, RRSAgent, tantek, dwhly, xmpp-social, timbl_, nightpool, bwn, trackbot, cwebber2, jungkees, wilkie, dlongley, MMN-o, rhiaro, jet, ben_thatmustbeme, # 17:23 Zakim ... csarven, jaywink, bigbluehat, sknebel, tsyesika, astronouth7303, Loqi, Gargron, puckipedia, raucao, tcit, ajordan, mattl, DenSchub, saranix, jankusanagi_, aaronpk, KjetilK_, # 17:24 ajordan tantek: okay, aaronpk, how are we doing with websub? # 17:24 ajordan ... I believe last time we met we discussed waiting until about now to collect impl. reports for trying to transition # 17:24 ajordan aaronpk: Google submitted an implementation report # 17:24 ajordan ... so pubsubhubbub.appspot.com now has a report in here # 17:24 ajordan ... it's their official hub and it passes all the tests # 17:25 ajordan ... they used the test suite to confirm everything # 17:25 ajordan ... this morning tantek snuck in a report for Falcon, as a publisher # 17:25 ajordan ... so we're doing pretty good on number of reports now # 17:25 ajordan tantek: great, I have this vague memory of Gargron doing a live impl. report in IRC for Mastodon # 17:26 ajordan aaronpk: good question, I don't remember the date but I do remember him braindumping a report into IRC # 17:26 ajordan ... just doing a checklist on IRC instead of GitHub # 17:26 ajordan ... once I find that I'll convert it into a proper report on GitHub # 17:26 ajordan tantek: was that selfreported or did he use the test suite? # 17:26 ajordan aaronpk: I believe selfreported cause he was doing it pretty quickly # 17:26 ajordan ... he thought it would be a quick thing, just fill out the report, but it wasn't # 17:27 ajordan tantek: we need to decide as a group if we're ready to ask for PR transition # 17:27 ajordan ... sandro, any updates on when it'd be good to start that ball rolling? # 17:27 cwebber2 one thing you may notice in the AP test suite is that it uses websockets and it disconnects you very quickly, lol... I need to implement websocket ping, oops. but that'll happen # 17:27 ajordan sandro: I think we're a while from the deadline but I don't see a reason to wait # 17:27 cwebber2 also eventually there will be a RESTful API that you won't have to use websockets for ;P # 17:28 ajordan tantek: I thought we were waiting for the Google report and the Mastodon report so assuming aaronpk can dig up that IRC report I think that satisfies the previous conditions we had set # 17:28 ajordan ... for past CR to PR transitions we've started a wiki page for the specs we had to do that for with a bunch of different items to make sure we'd done # 17:28 ajordan ... aaronpk you should start that, if you haven't already # 17:28 ajordan ... that's transition request I belive, it's got a bunch of fields you have to fill out/qs to answer # 17:29 ajordan ... let's open the floor to discussing if the group thinks we're ready to take WebSub to PR # 17:30 ajordan ... actually before I ask that, aaronpk are there any outstanding normative issues we need to resolve? # 17:30 ajordan aaronpk: there's a couple things open for discussion # 17:30 ajordan ... there's a note from tantek from this morning about clarifying the publisher relationship # 17:30 ajordan ... more explicitly documenting that there isn't anything specified # 17:31 ajordan tantek: so looking at this list... there's something about topics, is that a feature request? # 17:31 ajordan tantek: presumably that's for a future version? or are we trying to figure out how the spec could implement it as-is # 17:31 ajordan aaronpk: I think it's a new enough thing it'd be better as a future version # 17:31 ajordan ... especially if our goal is to keep things compliant with PuSH # 17:32 ajordan tantek: any other opinions, has anyone else had a chance to look at this issue? # 17:32 ajordan ... we'll I'll ask for a group resolution based on that # 17:32 tantek PROPOSED: Accept websub issue 110 as a future feature request, not for current version of WebSub. # 17:33 ajordan RESOLVED: Accept websub issue 110 as a future feature request, not for current version of WebSub. # 17:33 cwebber2 +0.5 (sounds good but i haven't thought about it much) # 17:33 ajordan tantek: aaronpk if you could update that issue commenting the resolution that'd be great # 17:34 ajordan ... looks like a doc issue rather than a change-the-spec issue # 17:34 ajordan aaronpk: it was a change-the-spec issue but at the end he said there wasn't a conflict like he thought # 17:34 Loqi [Alkarex] #106 Suggestion: Use HTTP 410 Gone # 17:34 ajordan tantek: that then takes us to 410 Gone? where did we end with that # 17:35 ajordan aaronpk: this resulted in a change that we added a while ago so we're just waiting for the commenter to confirm that then # 17:35 ajordan tantek: okay, so nothing outstanding for us to do then # 17:35 Loqi [dissolve] #118 add implementation report on diasporas behalf # 17:35 ajordan tantek: looks like the last thing we have in normative is the... oh! # 17:35 Loqi [aaronpk] #97 Discussion for Host-Meta feature At Risk # 17:35 ajordan ... left it open deliberately to try to see if anyone still cares # 17:36 ajordan ... we have to resolve this before going to PR, that's what the CR period is for # 17:36 ajordan aaronpk: right now there are a couple votes in support of dropping it, haven't seen much discussion in support of it except for Julien's last comment about leaving it # 17:36 ajordan tantek: did we get any hints from impl. feedback... did anyone check the "we implement this" box for host-meta? # 17:37 ajordan ... ironically we have the least amount of reports from subscribers, which is where this would come in # 17:37 ajordan tantek: well I think it would be publishers depending on it # 17:37 ajordan ... these three publishers do not publish at all on host-meta # 17:37 ajordan tantek: so not only do they not depend on it, they don't support it at all # 17:37 ajordan aaronpk: correct, none of the three have advertised the hub via host-meta # 17:38 ajordan ... not a surprise they don't implement it if they don't want # 17:38 ajordan aaronpk: the need for this would be if a publisher was *only* able to publish via host-meta # 17:38 ajordan ... that's the current argument for keeping it in the spec # 17:38 ajordan ... currently in the three reports none of them even advertise host-meta, much less depend on it # 17:39 ajordan sandro: well they wouldn't advertise it if they need it? # 17:39 ajordan ... can't rely on host-meta if no one supports it # 17:39 ajordan aaronpk: as a publisher you can advertise your publisher, _always_, with any of them # 17:40 ajordan sandro: only reason for host-meta is for publishers who can *only* do host-meta # 17:40 ajordan tantek: I thought for some reason diaspora did but the impl. report that ben_thatmustbeme just submitted doesn't have it? # 17:41 ajordan aaronpk: Mastodon uses webfinger for other things but not for hub discovery # 17:41 ajordan tantek: okay so people use host-meta for other things but not for websub # 17:41 ajordan tantek: okay, at this point there seems to be overwhelming evidence there's not even an indication of implementors # 17:42 ajordan sandro: I think there's evidence people would want it but there's no way to do it # 17:42 ajordan ... I think it would be a good thing but I think we should probably drop it # 17:42 ajordan tantek: I have a difference of opinion on experiments sandro but generally I agree with you # 17:42 aaronpk ben_thatmustbeme did you say something? definitely didn't hear you # 17:42 ajordan ... e.g. supporting random feed formats no one consumes yet has been known to happen # 17:42 ben_thatmustbeme doesn't pubsubhubbub support it, and so any existing subscribers from push 0.4 would have supported host-meta # 17:43 Loqi [aaronpk] #97 Discussion for Host-Meta feature At Risk # 17:44 Loqi [aaronpk] #97 Discussion for Host-Meta feature At Risk # 17:44 ajordan tantek: aaronpk go ahead and note the summary on the issue and make the edits to the spec # 17:44 ajordan tantek: ok, that takes us to... I think that's it for the normative issues on WebSub? # 17:45 ajordan ... are there any normative changes outside of issues you wanted to bring up? # 17:45 ajordan aaronpk: I don't think so, lemme check the draft thought # 17:45 ajordan tantek: cause I would expect the Director to do a diff # 17:45 ajordan aaronpk: yeah we've talked about all the changes we've done so far # 17:46 ajordan ... so the only new issue is the one I brought up # 17:46 ajordan ... I can ask Evan to chair for this one since I brought it up but it's informative # 17:46 ajordan ... but we can talk about it in case it has normative impact # 17:46 Loqi [tantek] #117 informative NOTE pub->hub protocol unspecified, and examples # 17:46 ajordan ... so, could you explain what the issue is tantek? # 17:47 ajordan tantek: if you look at seciton 6 it has one very short paragraph on what publishers should do # 17:47 ajordan ... I requested an informative note making it clear that the pub->hub protocol is left unspecified, and explicitly say what some public hubs have been doing # 17:48 ajordan ... which is to send a POST request with some well-known key names # 17:48 ajordan eprodrom: I think this is because we have previously-existing PuSH versions where this was their mechanism for pubisher notification right? # 17:48 ajordan aaronpk: what's happened with the spec is that it never specified how publishers verify hubs because some publishers integrate into hubs # 17:49 ajordan ... like when it's built into your blog, so you don't need an external API # 17:49 ajordan ... Superfeedr and Google happen to implement the same API because they're both external hubs # 17:49 ajordan ... those are the two links tantek dropped into the issue, for service docs # 17:49 ajordan ... it's sort of become a de facto standard because public hubs do it that way # 17:49 ajordan ... best thing we can do is say "this is what the situation is" # 17:49 ajordan ... but we can't make it required without breaking things # 17:50 ajordan eprodrom: can we mention a mechanism for doing things specific ways and what the parameters are # 17:50 ajordan ... even if we say this is unspecified but this is how you do it... is that a spec # 17:50 ajordan ... it seems simple and pretty clear is the question there # 17:50 ajordan cwebber2: I just noticed in some specs I've been reviewing they've been doing it's RECOMMENDED... instead of SHOULD or whatever you can do RECOMMENDED # 17:51 ajordan ... that seems like a good way to shove people in the right direction # 17:51 ajordan sandro: RECOMMENDED is defined as a synonym of SHOULD in RFC2119 # 17:51 ajordan aaronpk: I'd like ???, not sure if we can do that # 17:51 ajordan eprodrom: I'd like if it was normative, it seems pretty straightforward # 17:51 ajordan ... you'd need some kind of definition as to what that means # 17:52 aaronpk s/???/it to be a normative part of the spec if that situation applies to you/ # 17:52 ajordan ... my question is, we're in CR right now, we'd be adding an additional "module" to WebSub # 17:52 ajordan tantek: the reason I raised this issue was, I was filing my impl. report and going through all the steps # 17:52 ajordan ... and I realized as I got to the last step that what I was doing wasn't in the spec # 17:52 ajordan ... realized I was following docs from public hubs # 17:53 ajordan ... made sense to at least mention that that documentation exists # 17:53 aaronpk ben_thatmustbeme, both superfeedr and google use hub.url # 17:53 ajordan ... but trying to be conservative and not add normative text, instead just ack the current situation # 17:53 ajordan ... I would be _for_ normative text in a *future* WebSub version # 17:53 ben_thatmustbeme i also was working on a hub implementation that uses webmention instead of that methods # 17:53 ajordan ... if it gets more uptake and there are no objections... all the usual spec iteration stuff that involve the broader community, I could see it going into a 1.1 # 17:54 ajordan ... in particular I'd like Julien's opinion on any normative change # 17:54 ajordan ... I felt it was still valuable to include an informative short note, stating reality # 17:54 ajordan eprodrom: could I propose a second path? we could have a separate document # 17:54 ajordan ... that defines what the state of the world on pub to hub protocol and refer to that document from here # 17:54 ajordan ... the advantage being that could evolve separately # 17:55 ajordan aaronpk: I support the idea of a Note for specifying that relationship, I think tantek's right that the spec should say *something* # 17:55 ajordan ... "it's unspecified, for example here's one way you could do it" # 17:55 ajordan ... as a reader you're not missing something, it really is not here # 17:55 ajordan eprodrom: do we think we'll be further standardizing this interface? # 17:55 ajordan aaronpk: yeah as a separate future document or as a Note or something # 17:56 ben_thatmustbeme +1 to both a note inline as "this is unspecified, but here is one way that has been used" # 17:56 ajordan eprodrom: for me I'm wondering at one point do we begin the little branch/bud that will become pubhub # 17:56 ajordan ... do we note it here and link to it or just leave it as "two hubs do it this way" and that's it # 17:56 ajordan ... how we capture this kind of protocol, whether it's in a 1.1 or a new pubsub spec # 17:57 ajordan ... but I think that doesn't preclude us at least mentioning the state of where things are # 17:57 ajordan ... it makes reading the spec seem like there's something missing if we don't say anything here # 17:57 ajordan ... I'm trying to figure out the best way to handle that off # 17:58 ajordan aaronpk: I think I would prefer to add this text as-is instead of preemptively linking to something that doesn't exist yet # 17:58 ajordan ... let's plan on writing that but in the meantime this seems useful # 17:59 ajordan eprodrom: sounds good, aaronpk I assume you'll do some wordsmithing on here # 17:59 Loqi [tantek] #117 informative NOTE pub->hub protocol unspecified, and examples # 18:00 Loqi [tantek] #117 informative NOTE pub->hub protocol unspecified, and examples # 18:00 ajordan eprodrom: I'd like to hand the floor back to tantek unless there's something else you need? # 18:01 ajordan ... I believe that's enough to close all the normative issues on websub # 18:01 ajordan ... and to make sure that informative issues don't have any normative implications # 18:01 ajordan ... already talked about impl. reports, got enough for publishers and subscribers # 18:01 ajordan ... not counting Mastodon's since I haven't officially filed that one yet # 18:02 ajordan tantek: and the diaspora one that ben_thatmustbeme filed during the telecon # 18:02 ajordan aaronpk: yeah the diaspora one too, that's a publisher. that makes 3 publishers now # 18:03 ajordan aaronpk: sorry, got that mixed up - 5 hubs, 3 publishers, and now 3 subscribers with... I can't count # 18:03 ajordan tantek: did Mastodon submit as a publisher? as a hub? # 18:04 tantek PROPOSED: with edits agreed during telcon, transition WebSub from CR to PR, with aaronpk writing up transition request on the wiki # 18:04 ajordan if someone could scribe after we're done with this topic that would be great # 18:05 ajordan RESOLVED: with edits agreed during telcon, transition WebSub from CR to PR, with aaronpk writing up transition request on the wiki # 18:06 ajordan tantek: congrats everyone, this is a pretty big transition # 18:06 sandro happy to be on vacation for the next two weeks :-) # 18:07 rhiaro may have made the best pizza she's ever made. Thanks for being there everyone. Four kinds of vegan cheese is a definite help. # 18:08 ben_thatmustbeme ajordan: i have a bunch of other changes queued up on paper, and I don't think i have any normative changes # 18:10 ben_thatmustbeme rhiaro: i don't think i have made any changes since the last update to document status # 18:11 ben_thatmustbeme cwebber2: we have a call tomorrow, i also had a call with another group that is using AS in a really large way, and they are joining the CG # 18:12 ben_thatmustbeme ... with the transition of websub to PR, we are just waiting on the progress on AP and getting that to PR # 18:13 Loqi ben_thatmustbeme has 82 karma in this channel (251 overall) # 18:14 tantek 17 min early from our extended telcon period! # 18:14 Loqi ajordan has 16 karma in this channel (17 overall) # 18:14 ajordan aaronpk: lemme look up the phone number thing for you # 18:14 Zakim As of this point the attendees have been ben_thatmustbeme, rhiaro, tantek, aaronpk, ajordan, sandro, tsyesika, cwebber, eprodrom # 18:14 Loqi ajordan has 17 karma in this channel (18 overall) # 18:14 Loqi ben_thatmustbeme has 83 karma in this channel (252 overall) # 18:21 ajordan ben_thatmustbeme: thanks for posting the minutes, I totally forgot # 18:28 ajordan has always thought it was a huge bummer that w3cmemes.tumblr.com does not take submissions # 18:29 tantek ajordan: they may be open to it - I think you just need to drop suggestion in the places # 18:30 ajordan hahaha I remember reading that one but that was before I was involved here # 18:30 tantek ajordan: yeah that was from CSSWG meeting a while ago # 18:30 Loqi [Gargron] mastodon is publisher, hub and subscriber # 18:31 Loqi ben_thatmustbeme has 84 karma in this channel (253 overall) # 18:55 puckipedia Gargron: ooh, Kroeg seems to still be fine with the new activitypub structures # 18:55 puckipedia now to actually test the federation :D # 19:03 cwebber2 works on setting up the SocialCG call page for tomorrow # 19:05 rhiaro works on documenting her pizza for social media # 19:05 cwebber2 rhiaro: you can't have a social media network without food pics # 19:14 puckipedia I feel like AP is missing some more server->client signaling thingies imo # 19:14 puckipedia like, things that aren't /required/ in the spec, but maybe as extensions # 19:14 cwebber2 puckipedia: you mean like notifying your client that events have occurred without polling? # 19:15 puckipedia I mean more stuff like filtering your inbox on object types, and something (like I already have :P) for relevant entities # 19:15 puckipedia so a client can easily and cheaply say "you have 4 unanswered follow requests" # 19:15 cwebber2 puckipedia: ah yeah, so we've talked about that, but agreed that it should be an extension # 19:15 puckipedia I have a relevantEntities thingy at least # 19:16 tantek I didn't know four kinds of vegan cheese even existed! Thought it was just daiya (which is kinda ugh IMO) # 19:16 puckipedia . o O ( an extension on the ActivityPub Follow to indicate it must be followed back ) # 19:16 puckipedia for implementing mutual-follow structures lol # 19:18 cwebber2 AP doesn't specify it but you could see how it could easily be done as an extension (one that wouldn't even need to add vocaublary) # 19:20 tantek suggests cwebber2 write-up "follow-back" as an AS2 extension for 2018-04-01 # 19:22 rhiaro tantek: oops the urls in alt attrs must be a bug in my photo album code # 19:23 rhiaro and yeah Tesco in the UK have had a whole new range of vegan cheeses for a while, so when I went back for a few days last week I basically packed nothing and returned with an entire backpack full of not-cheeses # 19:24 rhiaro I like daiya, but I think these might even be better # 19:24 tantek I've tried to like daiya several times and can't even # 19:24 rhiaro which one though? there's a great jalapeno chedder # 19:24 rhiaro the daiya solid block cheese are better than the grated ones # 19:25 tantek cashew cheese works really good for spreading, like a cream cheese substitute # 19:25 cwebber2 cashew blue cheese that legit has the appropriate mold in it # 19:26 cwebber2 the internet says it's great, but it takes a month to make and I'm not that patient # 19:26 rhiaro cwebber2: oooh nice. One of the Tesco ones is blue cheese but I don't know if it's legit mold # 19:26 rhiaro I have a recipe for fermented cashew cheese that only takes a couple of days and has two ingredients so I can't wait to try # 19:26 aaronpk rhiaro: whoa! I wonder who makes it and if there is an equivalent here # 19:27 rhiaro There's a jalapeno one of these too, which is not pictured, by my favourite so far # 19:28 rhiaro well, that was a nice interlude from all the usual techno babble :p (sorry for interrupting serious AP conversation) # 19:30 Loqi rhiaro has 157 karma in this channel (276 overall) # 19:33 ajordan puckipedia: FWIW better inbox querying/filtering is already on the list of extensions to explore # 19:35 ajordan also I like how tantek's first instinct upon seeing rhiaro's pizza was "ooooo lemme check out the HTML source!" # 19:37 ajordan also, about finding a way to indicate on Follows that it doesn't have to be ack'd, I thought that defeated half the point? which was that a Follow is one of the most critical types of Activities and we _really_ want to make sure that they make it over the wire? # 19:38 Loqi rhiaro has 158 karma in this channel (277 overall) # 19:42 rhiaro ajordan: there's making it over the wire, then there's it being ackd by the human at the other side # 19:43 ajordan rhiaro: well if the human on the other side doesn't have a "private" account (meaning, an account where Follows have to be ackd) the impl. autoacks # 19:43 rhiaro the implementation *may* auto-ack but it's not required # 19:44 ajordan I thought we resolved to make it required always? # 19:44 ajordan uhhh I may be totally misremembering, lemme look up the issue # 19:45 rhiaro oh, we did decide the reject was mandatory too # 19:45 rhiaro oh right and then I wrote a paragraph about why that concerned me # 19:51 ajordan okay this *may* be totally stupid, so feel free to say so :P # 19:52 ajordan what if we made it mandatory to send an Accept/Reject right away # 19:52 ajordan so in a "public" account, the server would send an Accept immediately # 19:53 ajordan and in an "approvals-only" account, the server would send a Reject and if the user approved the follow, the server would send an Accept (or whatever would undo the Reject it'd already sent) # 19:54 ajordan so there you get acking, and you get a strong signal as to whether the user being followed has to approve your request or not # 19:55 ajordan ALTHOUGH this is a bit of an abuse of Reject and it also occurs to me that you would still have to know whether follows require approval *before* actually sending a Follow and getting Accept/Reject back # 19:55 ajordan feels bad for returning to the techno-babble after the far more interesting food discussions :P # 19:55 rhiaro I feel like an immediate reject might be a kind of confusing way to handle it # 19:56 puckipedia ajordan: I'm not really positive about this; mostly due to the fact that it is, as you say, an abuse of a reject; and also would it mean that ActivityPub has to specify delivery timeframes? # 19:56 rhiaro What do you lose by just never sending a reject? (or sitting on it for ages?) # 19:56 rhiaro it's only impactful if you want to Accept but don't send anything. Which is kind of your problem # 19:58 cwebber2 maybe it should be, once a decision is made to accept or reject a follow request, the server MUST return the corresponding accept/reject # 19:58 puckipedia it means it might show up in "pending follows/following" # 19:58 cwebber2 in effect, that models someone who gets a follow on a system that has manual accept/reject and never bothers to reply # 19:58 rhiaro how it's displayed to the human is up to the implementation # 19:58 rhiaro yeah I sit on facebook friend requests for years without clicking # 19:59 rhiaro I think I eventually covered that in my reply on the issue # 19:59 puckipedia well, I also consider a Reject on any Follow an immediate block on future Accepts # 19:59 rhiaro like you can't expect to get stuff pushed to you unless you send an Accept # 19:59 puckipedia so you can only either Accept and then Reject a Follow, or Reject it immediately # 19:59 rhiaro I'm so full of pizza I'm not expressing things right # 20:00 rhiaro ..read what I wrote when I wasn't full of pizza :) # 20:01 rhiaro but maybe now is a good time to put my AS2 serialiser fixes into production # 20:01 ajordan rhiaro: you lose the acking of a Follow being received # 20:01 ajordan generally though I'm also -1 on my proposal, for the Reject abuse and stuff # 20:05 cwebber2 well I have dustycloud.com but I'm not doing anything with it :P # 20:05 ajordan uhh... *shifty eyes* that is definitely, 100% what I said ;) # 20:05 rhiaro argh okay I think the responses are finally okay if you request anything on my site as activity+json # 20:06 rhiaro I had to beat up the json-ld serialiser to do this # 20:06 rhiaro I should probably check they're still valid json-ld tbh # 20:16 ajordan aaronpk: forgot I was going to check the phone number thing for you # 20:17 ajordan I don't know it that'll work with hangouts but it works with my stock Android dialer # 20:17 ajordan I think I found that on the wiki somewhere but I'm not sure where # 20:22 aaronpk lol the hangouts app doesn't let me edit my contacts # 20:25 Zakim excuses himself; his presence no longer seems to be needed # 20:26 Gargron puckipedia: soooooo whaddup, what's working? # 20:26 puckipedia sorry, went to fix some small extensions with my server # 20:26 puckipedia and then a large important thing Occured # 20:27 Gargron im frustrated over the whole problem space of upgrading the whole network to another protocol, especially in how it relates to disseminating OStatus-originating content via AP # 20:28 Gargron some of the origin servers will upgrade and start supporting AP, and then you could somehow try to update some of the data stored on your remote servers to reflect that, and disseminate that # 20:28 Gargron but some will never upgrade # 20:28 Gargron some are gnu social or friendica # 20:30 Gargron i just dont get what i need to do # 20:31 Gargron currently since we just take whatever's in the DB and serialize it as activity+json, it means ostatus-posts are mostly identical to AP posts (because the db schema is obviously the same, and normalized), but the URI format differs # 20:31 Gargron atom URIs are like tag:example.com;123;222 # 20:32 Gargron when we're talking about local posts, it's fine - we generate URIs on the fly, so we can generate the correct URI and put it in "id", and generate the ostatus one and put it into "_:atomUri" for cross-referencing # 20:33 Gargron but when we want to, say, "announce" a remote post, the URI recorded in the db is the only thing we have # 20:33 Gargron and it happens to be tag:stuff # 20:33 Gargron so that goes into the "id" field # 20:35 Gargron that obviously messes some algos up, when you want to resolve something, or even if you wanna check "is this *my* URI" because we're expecting a different format there # 20:50 jaywink Gargron: do you plan on keeping OStatus support or slowly ramping it down? # 20:51 Gargron slowly ramping it down # 20:51 jaywink I'm going to have fun hacking AP support to Socialhome at some point, since now I support Diaspora protocol and it's completely different :P and I want to keep both # 20:53 jaywink comments/likes might be a bit tricky/impossible, but can't have everything # 20:53 jaywink I mean relaying for example comment in AP to diaspora users # 20:54 jaywink hubzilla I think sends them as quoted new comments by the original author, which is kind of dirty but works - but of course you can't verify what the commenter *actually* said, you need to just take the word of the content owner # 20:55 jaywink relaying comments from diaspora to AP would probably require something similar too - so will probably go that way bwn joined the channel
# 21:59 puckipedia Gargron: hmm, I'd like to see all the posts slowly moved to use new ActivityPub IDs (even old OStatus posts from Mastodon) - then adding a flag to enable/disable OStatus support # 22:05 puckipedia Gargron: current status # 22:05 puckipedia the Follow Activity towards Kroeg 500'd # 22:05 puckipedia ... in the template service # 22:06 puckipedia ehm, websubwise # 22:07 puckipedia wait no, this is correct I think # 22:07 puckipedia websub is a middleware for unknown reasons # 22:12 puckipedia Gargron: so the issue is probably that I'm incredibly lazy and allow someone to e.g. POST an application/activity+json into the inbox, but Accept: application/atom+xml to get OStatus back # 22:13 puckipedia and my guess is that Mastodon isn't doing an Accept: on POST requests # 22:22 puckipedia good news though # 22:22 puckipedia ttp://localhost:3000/users/admin", "object": "https://lol.puckipedia.com/users/puckipedia", "@context": "https://www.w3.org/ns/ac # 22:22 puckipedia tivitystreams"} # 22:23 puckipedia less good news though: ... there's no audience, and I'm pretty sure I don't have the weird filtering on my inbox because Reasons # 22:29 Gargron accept: on POST? wut? # 22:30 puckipedia eh, that's a tiny thing I have fixed on my side # 22:30 puckipedia the bigger question is: the Follow has no audience # 22:31 puckipedia because it is sent to someone, and people may (or may not) be allowed to see it? # 22:31 Gargron i suppose i could put that field in but seems absolutely redundant, no? # 22:31 puckipedia each ActivityPub Activity has an audience # 22:32 puckipedia e.g. I could say "I'm now following cwebber2" to cwebber2, and also my existing followers # 22:32 puckipedia like, tbh, it doesn't make sense to keep the Follow private, as there's also a usually-public list of who you follow # 22:33 puckipedia if I get an object with 0 audience items I should probably just make it public # 22:33 cwebber2 personally I don't use the latter property in my implementation, and activitypub doesn't require it # 22:33 puckipedia I consider AP "audience" to be to/cc/bto/bcc/audience taken together # 22:34 cwebber2 not totally paying attention to this conversation, maybe should read backwards # 22:34 puckipedia cwebber2: basically, Mastodon doesn't send an audience on Follow # 22:35 cwebber2 you probably want to if you're delivering it, probably to the object # 22:35 puckipedia yeah; I'd just say "to": "thePersonYou'reFollowing" # 22:35 cwebber2 of course you won't know what bto/bcc is on the receiveing side of federation anyway ;D # 22:36 puckipedia . o O ( bto: followers collection ) # 22:36 cwebber2 this really only seems like a big worry though in the case that you're using sharedInbox # 22:36 puckipedia I'll try to send it to the sharedInbox lol # 22:41 puckipedia yup I get the proper follows now! # 22:42 puckipedia also woops I can't serialize booleans properly lol? # 22:42 puckipedia I think I forgot booleans existed # 22:53 Gargron im gonna reboot into linux, so might rejoin as eugr and not have any channel history # 22:56 puckipedia woops entity.updated is set to 0001-01-01 # 23:15 puckipedia oh, keyId in the signature is the owner, not the key's ID? # 23:16 puckipedia Gargron: ??????? # 23:21 puckipedia " MASTODON HACK REMOVE ME BEFORE COMMITTING" # 23:21 puckipedia going to quickly check if this fixes the bug I had :P # 23:32 Loqi [puckipedia] #4606 ActivityPub: Key signature verifying only works when key ID is a fragment of the user # 23:34 puckipedia made a quick workaround for both # 23:34 puckipedia because I want to know if it works with Kroeg also I want to go to bed # 23:40 puckipedia actually I have more questions # 23:42 puckipedia ?? I think there was just a typo made in this code?
xmpp-social [ajordan] Ayyyyyyyy
rhiaro fortunately we have a record of these things https://www.w3.org/wiki/Socialwg/2017-08-01-minutes#Summer_meeting_schedule
cwebber2 Gargron: maybe this could go on the socialcg tracker to discuss? https://github.com/swicg/general/issues
jaywink I think the way diaspora does the "a node died, we should probably not do timeouting requests forever" is to keep a last success timestamp. If the node doesn't have any successes for X days -> it is not posted to any more. But no actual data is changed, it's just a server implementation thing, not protocol. This is also how my relay system works.
RRSAgent logging to http://www.w3.org/2017/08/15-social-irc
ben_thatmustbeme present+
tantek PROPOSED: approve https://www.w3.org/wiki/Socialwg/2017-08-01-minutes minutes
ajordan RESOLVED: PROPOSED: approve https://www.w3.org/wiki/Socialwg/2017-08-01-minutes minutes
Loqi Tantekelik made 1 edit to [[Socialwg/2017-08-15]] https://www.w3.org/wiki/index.php?diff=104065&oldid=104039
@rhiaro During today's @SocialWebWG call I made the best pizza I've ever made. It was helped by four kinds of vegan ch [...… https://twitter.com/i/web/status/897537222978789376 (twitter.com/_/status/897537222978789376)