From W3C Wiki
Jump to: navigation, search

11:08 < cwebber2> ========== START MEETING LOGGING ==========

11:08 < cwebber2> scribe: nightpool

11:08 < cwebber2> present+

11:08 < eprodrom> present+

11:08 < nightpool> present+

11:08 < yookoala> +1

11:09 < eprodrom> q+

11:09 < yookoala> present+

11:09 < cwebber2> present+ ajordan

11:09 eprodrom: I wanted to ask, could you coordinate with aaron and get us back on the schedule here, so we don't have to do this again?

11:09 < eprodrom> thanks!

11:09 cwebber2: yep, i'll set up a todo item for myself right now

11:10 < cwebber2>

11:10 * ajordan is here

11:11 < cwebber2> #topic AS2 editor's draft

11:11 cwebber2: three items today, two of them about AS2 extensions and 1 about version control hosting (somewhat in re: the github acquisition)

11:11 cwebber2: as requested, we'll timebox the first 2 issues for 30 minutes so we have 30 minutes for the last item

11:12 eprodrom: in our last meeting we talked about making an errata as our main form of updates, but we mentioned updating an "editor's draft" as well so I wanted to talk about that a little.

11:13 < cwebber2> q+

11:13 < ajordan> q+

11:13 eprodrom: my main goal for this is that someone looking at all the published updates + the errata, they would have all the information. So I think the editor's draft would be useful as a document applying the errata to the published document.

11:13 < nightpool> .. (And then maybe link to it from the errata).

11:14 < cwebber2>

11:14 < mayel> present (lurking)

11:14 cwebber2: +1, although I want to make the comment that we're not making a editor's draft (since we're not a WG), we're making a "community group report".

11:14 ajordan: so just to clarify, we're just discussing only adding stuff in errata, right? New features, other stuff we're adding, wouldn't end up in this document, right?

11:15 * cwebber2 welcome mayel

11:15 eprodrom: Yeah, the worry is that it'll get confusing very fast. If we're saying 'you should read this because it's an updated version', that's not the place for us to be adding experimental features etc.

11:16 < eprodrom> PROPOSED: the CG will maintain an "editor's draft" of AS2, applying only the existing errata changes to the final REC version of AS2

11:16 < cwebber2> +1

11:16 < nightpool> +1

11:16 < ajordan> +1

11:16 < emacsen> present (lurking)

11:16 < yookoala> Should I vote too?

11:16 < eprodrom> +1

11:16 < yookoala> +1

11:16 < yookoala> Thanks.

11:17 < cwebber2> RESOLVED: the CG will maintain an "editor's draft" of AS2, applying only the existing errata changes to the final REC version of AS2

11:17 cwebber2: Yep, all members of the community group are welcome to vote.

11:17 < eprodrom> That is great!

11:17 cwebber2: there are a couple new people here, so let's do a quick round of introductions. ajordan can you go first?

11:18 (ajordan: can you scribe your intro for the IRC people? i missed it :( )

11:18 * ajordan nightpool sure just a sec

11:19 < cwebber2> nightpool: I'm nightpool I do mastodon development and I run a mastodon server

11:19 < eprodrom> whoa, good domain

11:19 * ajordan yookoala FWIW quick crash course in W3C voting: +1 = enthusiastic yes, +0 = yes I suppose, -0 = I don't like it but I wouldn't block anything over it, -1 = strong objection, I would not be okay moving forward with this proposal

11:20 < cwebber2> yookoala: I'm koala, I'm a php developer and I'm a user of mastodon, also do some translation work for Mastodon, which is how I learned about ActivityPub. Trying to do some work on it with activypub and federi(ze?) with mastodon. I think I have an issue with geojs and proposed to use activitypub with geojs instances. I recently checked the issue and found so many people interested in federation and in activitypub somehow

11:20 < cwebber2> yookoala: so I fired up issues on different repos

11:20 < eprodrom> yookoala++

11:20 < Loqi> yookoala has 1 karma

11:20 < eprodrom> Awesome!

11:20 < nightpool> yookoala is also the author of which is a proposal for git federation

11:20 < Loqi> [git-federation] gitpub: An extenstion to ActivityPub for web-based Git services federation.

11:20 yookoala: I also created the repository where I drafted how a federation protocol works and someone else gathered that repository, I think I'm simply here to learn more about activitypub and see how we might extend the standard to work with git federation

11:21 < nightpool> thanks chris!

11:21 < cwebber2>

11:21 ajordan: I'm ajordan, I maintain which eprodrom wrote originally, and I'm involved in the IndieWeb community as well. also I comaintain a privacy site called PRISM Break which basically lists services vulnerable to NSA surveillance programs and then free alternatives, which are usually decentralized

11:22 eprodrom: I think a couple weeks ago we had an open issue on documenting our extension process for AS2. In particular, we have this big context document with multiple namespaces in it. So there's a question of how to include new namespaces or extension to this document

11:22 ... and it probably would not be out of line to start discussing this in terms of git federation.

11:23 eprodrom: in this document I covered four types of extensions we might see in as2. quickly: 1. established vocabularies. These are vocabs that have gone through some sort of standards process that are frequently used as a extension of as2

11:23 < cwebber2> q+ to say I'd probably prefer the established vocabularies not in general go into the as2 context, but advertising where to find them is great

11:23 ... so the idea is that we'll have a wiki page saying "when people are adding licensing information, they'll often use the cc schema'

11:23 < aaronpk> Not on the call and can't really participate today but wanted to introduce myself anyway. I'm one of the co-chairs of this group, and also the editor of a few specs, including IndieAuth, a federated identity and authorization protocol.

11:24 ... depending on their needs, we may add these vocbularies to the as2 context document. I think that would be up to the CG to decide and we'll start github issues to decide that.

11:24 * cwebber2 hello aaronpk! :)

11:25 * ajordan hey aaronpk, nice to see you!

11:25 * ajordan that reminds me cwebber2, I'm about to update the agenda

11:26 * ajordan just as a heads up :)

11:25 eprodrom: Second would be single-implementation extensions, stuff that's pretty specific to a single implementation. We might want to include that on the extension page. If you're looking at the page, you might see this internal vocab. (this is the database id for this object etc). I think it's unlikely we'd want to expand lots of effort to put these through a formal process but it would be nice to document

11:26 ... the third kind would be multi-implementation extensions still in development. For example, i talked a couple months ago about using ap for dating applications. This would be a kind of way for us to declare a 'dating' namespace and start experimenting with that in a way we can all understand might be standardized at some point

11:27 ... the idea would be to do those as a sub-namespace of the as2 namespace. I'm not enough of a namespacing ninja to say exactly how to do that, but there's an example on the document of what might work.

11:28 ... and then the fourth would be stable multi-implementation extensions. If the CG decides, we would then include them into the main as2 document. So then you could use them with a prefix, without a prefix, whatever. Most of what we would be doing would be maintaining the extension wiki document, and then carving up namespaces/sub-namespaces.

11:28 < eprodrom> nightpool++

11:28 < Loqi> nightpool has 11 karma

11:28 < eprodrom> Thank you for scrolling

11:29 < ajordan> q+

11:29 < ajordan> q-

11:29 cwebber2: that all sounds pretty great although you sounded somewhat hesitent about putting established vocabularies into the main as2 document. I'm also somewhat hesitant to that as well, but do you think there's a possible compelling usecase for that?

11:30 eprodrom: handingling complex @context's is kind of a pain. If we're all using the same one, say security or whatever, it would be great to drop that.

11:30 cwebber2: makes sense.

11:31 nightpool: Question, there was concerns earlier around mutable contexts breaking signatures, what was the thinking around that?

11:31 < nightpool> :D

11:31 * ajordan scribin is hard, thx nightpool

11:32 eprodrom: agree, and I think it's a high bar to pass to include something in the main as2 document, so we'll have to be mindful of that.

11:33 ... can we get this on the agenda for next week and get a vote on it? I don't want to drop it going forward.

11:33 cwebber2: sounds great, i'll do that.

11:33 < eprodrom> \o/

11:33 TOPIC: Federating git + other vcs hosting

11:33 < cwebber2>

11:33 < Loqi> [git-federation] gitpub: An extenstion to ActivityPub for web-based Git services federation.

11:33 cwebber2: Next topic on the agenda is federating git and other version control hosting.

11:33 < cwebber2>

11:33 < Loqi> [21stio] #2 ask w3c/activitypub how to extend the activitypub protocol

11:34 < cwebber2>

11:34 cwebber2: there's lot of discussion on this issue, and lots of other issues, I'll link in some other places people are discussing it.

11:34 < cwebber2>

11:34 < Loqi> [cjslep] Please do not hesitate to reach out to me on Mastodon for any questions/concerns/comments surrounding the library that @Skehmatics mentioned. I obviously have a vested interest in the outcome of the decision, but wo...

11:34 < eprodrom> q+


11:34 < cwebber2>

11:34 < Loqi> [cjslep] Please do not hesitate to reach out to me on Mastodon for any questions/concerns/comments surrounding the library that @jas99 mentioned. I obviously have a vested interest in the outcome of the decision, but would h...

11:35 eprodrom: So I have one comment and one suggestion. My comment is that I'm excited to see lots of discussion about this--it's been a problem with git hosting for quite a while and i'm excited to see an effort to get it going.

11:35 ... I wonder if there's an organizational umbrella we can extend to the various git hosting projects, and say 'Hey, why don't you negotiate this out within the w3c?' Why shouldn't we make this a w3c issue?

11:35 < mayel> agree with @nightpool’s comment, I’ve been thinking this needs to happen for a while!

11:36 cwebber2: Do you think this group is the right place or should we set up another thing?

11:36 < eprodrom> push harder, cwebber2

11:36 eprodrom: I think that having a new git federation cg would be kind of cool

11:36 cwebber2: that sounds fine to me but we'd have to find someone that's interested in running that

11:36 < ajordan> q+

11:37 ajordan: i'm not opposed generally to having a new community group, but why wouldn't we do this within the existing community group. Is there a compelling reason?

11:37 < eprodrom> q+

11:38 cwebber2: The main question I guess is that we might not be able to prioritize it enough. Personally, I think it would be good not to lose the activitypub/as2 expertise and momentum that we already have in this group by moving to another group.

11:38 < yookoala>

11:38 < Loqi> [cjslep] #5 Meta: Self-Assembling Organization

11:38 < yookoala> q+

11:38 < yookoala> Thanks

11:38 * ajordan yookoala normally a bot named Zakim handles the queue but it's been misconfigured for a month or so now

11:39 eprodrom: I think it makes a lot of sense for the current group to accellerate that current process, but if there are a lot of people working on it they may outgrow the size of this group.

11:39 * ajordan yookoala should help you out

11:39 .... I wonder if there's a possibility of inviting/supporting a summit meeting between gogs/gitlab/gitea? That might be a great way to get started with this effort

11:40 < ajordan> s/accellerate/accelerate/

11:40 yookoala: I think there are many people who are interested in starting this workgroup, as linked above

11:40 cwebber2: maybe we should respond there asking people to come into this community group?

11:41 < tantek> good morning #social

11:41 cwebber2: Let's follow up on that thread, and see if people want to set up a special meeting, or join the biweekly metting, or something similar.

11:41 < tantek> I can't seem to make the mumble work with dustybutt . org

11:41 < nightpool> s/cloud/butt strikes again?

11:41 < ajordan> ahaha tantek do you use Cloud to Butt too?

11:41 < eprodrom> ha ha ha

11:42 < tantek> no so someone who does must have copy/pasta'd

11:42 cwebber2: So I think to lay this out a little, ??? laid out a way similar to the way I was thinking about this, where Users would be actors and Repositories would be actors, and they would be sharing activities that are happening in the repositories to actors

11:42 < nightpool> q+

11:42 cwebber2: does that sound right?

11:42 yookoala: that was how I was imagining it also

11:43 < tantek> it's definitely in the wiki!

11:43 < eprodrom> Yes

11:43 < tantek> lol

11:44 < nightpool> (i'm not scribing the cloud2butt discussion....)

11:44 < eprodrom> THAT IS AWESOME

11:44 < nightpool> see also

11:45 < eprodrom> We need to add a cloud-to-butt buster script to the wiki software

11:45 < ajordan> ok I'm going to go ahead and NOT try to fix the wiki anymore because I have it installed

11:45 * ajordan deep sigh

11:45 < tantek> q+ for having experimenting with federating github posts from his own site since mid February!

11:45 < ajordan> eprodrom++

11:45 < Loqi> eprodrom has 55 karma in this channel (56 overall)

11:45 < tantek> q+ to note he has been experimenting with federating github posts from his own site since mid February!

11:45 < tantek> q?

11:46 < cwebber2> nightpool: I just wanted to say that the first thing I'd like to see is a use of use cases we'd see from federated git hosting. One of the most important ones as far as I can see is a shared login system for filing issues and etc, and I don't think that existing activitypub systems think about that

11:46 < tantek> q+ to note he has been experimenting with federating github posts from his own site since mid February!

11:46 * Zakim sees tantek on the speaker queue

11:46 * ajordan oh nice tantek

11:46 * ajordan we had been manually keeping track of the queue because trackbot is broken

11:46 cwebber2: yeah, authentication and authorization is a pretty important issue here--especially because activitypub doens't specify much about that

11:47 < eprodrom> glad to be here

11:47 cwebber2: Also the ACL issue makes this a little more complicated since most people have been using an implicit ACL based on hosting

11:47 < tantek> present+

11:47 < yookoala> q+

11:47 * Zakim sees tantek, yookoala on the speaker queue

11:48 cwebber2: so, we could have a federated ACL type system, but I think that's very dangerous--synchronizing ACLs makes it easy to miss messages, etc.

11:48 < tantek> +1 to that

11:48 cwebber2: I see two main approaches other then that, both of them are effectively capabilities

11:49 .... you have oauth2.0 bearer tokens, which you can consider bearer capabilities.

11:49 .... You would have as some sort of HTTP header some token that permits you to do something

11:49 * cwebber2 oh I missed tantek

11:49 * cwebber2 tantek you will be next... the queue is broken

11:49 * tantek no prob cwebber2

11:49 < tantek> q?

11:49 * Zakim sees tantek, yookoala on the speaker queue

11:49 < tantek> ack yookoala

11:49 * Zakim sees tantek on the speaker queue

11:50 * ajordan we could also build a Wikipedia-style free-for-all where anyone can do anything

11:50 * cwebber2 oh it is working?

11:50 < nightpool> And then you also have OCAP-LD, but that specification is fairly new/in-progress and i have to disclaim that i'm also the editor on it

11:50 < cwebber2> q?

11:50 * Zakim sees tantek on the speaker queue

11:50 * tantek cwebber2 I had to invite Zakim

11:50 * ajordan I came here to make outlandish suggestions and chew gum and I'm all out of gum

11:50 * tantek cwebber2 if q doesn't work, then /invite Zakim #social

11:50 < cwebber2> ack tantek

11:50 < Zakim> tantek, you wanted to note he has been experimenting with federating github posts from his own site since mid February!

11:50 * Zakim sees no one on the speaker queue

11:51 yookoala: i'm not sure I see the usecase for a ACL type system--i'm mostly thinking of a notification type system for issues/PRs, and that's what I see the main usecase to be. (Although i'm also interested in learning more and open to other ideas)

11:51 < mayel> @ajordan that’s what git-ssb has been doing:

11:51 < cwebber2> nightpool: I think the question is more what is the use case for authorization and authentication

11:52 < cwebber2> nightpool: main use case is PRs and isssues

11:52 < yookoala> Thanks. Thats my question.

11:52 cwebber2: my bringing up ACLs was as a lets-not-do-that sort of thing

11:52 < ajordan> mayel: lol I meant it as a joke :P

11:53 < eprodrom> q+

11:53 * Zakim sees eprodrom on the speaker queue

11:53 < tantek> I can't hear anything

11:54 < eprodrom> cwebber2: you're glitching

11:54 cwebber2: I think the main usecase would be for someone who's authorized to write to the repository can merge/etc a PR. And people also have features where they close an issue to make it only writable by contributors

11:54 < yookoala>

11:54 < yookoala> q+

11:54 * Zakim sees eprodrom, yookoala on the speaker queue

11:54 cwebber2: And other questions like, "can you discover someone's SSH key from their username" and stuff like that

11:55 < tantek> q+ to note experimenting with federating github posts from his own site since mid February! (issues, comments, reacji)

11:55 * Zakim sees eprodrom, yookoala, tantek on the speaker queue

11:55 < cwebber2> q?

11:55 * Zakim sees eprodrom, yookoala, tantek on the speaker queue

11:55 < cwebber2> ack eprodrom

11:55 * Zakim sees yookoala, tantek on the speaker queue

11:55 < tantek> I still can't hear anything :(

11:55 < tantek> going to try hanging up and reconnecting

11:55 < tantek> nope still quiet

11:55 < ajordan> tantek: audio issues are resolved for me, I think it's your client

11:55 < tantek> ok I can hear now

11:55 eprodrom: I wonder if we can set up bounds like 'every repository has local access on a single server' and remote access would just be submitting issues, opening PRs, etc. I think those things would be the 80% solution and it would be great to get those in.

11:56 < cwebber2> ack yookoala

11:56 * Zakim sees tantek on the speaker queue

11:56 * nightpool -ish

11:56 < nightpool> +1 to eprodrom

11:56 < ajordan> eprodrom++

11:56 yookoala: I was thinking of a much simpler thing, and in my draft i was thinking of only doing the federation protocol--everything that happens within a git service happens inside the server, and that doesn't federate

11:57 < cwebber2> ah, I didn't see until now... it was in a branch

11:57 ... if someone is doing a pull request, that code is happening on my server only and i'm only sending a notification that i'm doing something with your code.

11:57 < cwebber2> q?

11:57 * Zakim sees tantek on the speaker queue

11:57 < cwebber2> q+

11:57 * Zakim sees tantek, cwebber on the speaker queue

11:57 ... With a pull request, i'm just sending a notification that something is happening to your repository and that I have a branch you can do something with

11:58 < cwebber2> ack tantek

11:58 < Zakim> tantek, you wanted to note experimenting with federating github posts from his own site since mid February! (issues, comments, reacji)

11:58 ... I initially hadn't thought about permission control and like the last person said, I think that if we can get it working without that would be a big win

11:58 * Zakim sees cwebber on the speaker queue

11:58 < ajordan> q+

11:58 * Zakim sees cwebber, ajordan on the speaker queue

11:58 tantek: I'm definitely interested in getting parts of this working in a federated way, with issues and comments and all that. I don't remember why I started working on this but since feburary, every post I've made on github has been on my own website and federated to Github using webmention

11:59 < tantek>

11:59 ... I've posted issue on my site which have then federated to github, I've posted comments issues on my site that have federated to github, I've posted replies to comments (slightly different case) and reactions from my own site to github

11:59 < yookoala> tantek: Thanks.

12:00 ... part of the idea would be to see how much of github's silo'd content can be done from my own site. It definitely brought up an issue like 'Who is allowed to close an issue?'

12:00 ... I'm definitely biased to say something like "The author of an issue is the only person allowed to close an issue" and i think there's a lot of UX-space to explore here

12:00 < cwebber2> ack cwebber

12:00 * Zakim sees ajordan on the speaker queue

12:00 < eprodrom> tantek++

12:00 < Loqi> tantek has 79 karma in this channel (441 overall)

12:00 < eprodrom> Great work!

12:01 ... and the possibility for a federated model of issue tracking to be very different from the existing silos.

12:01 < tantek> s/???/replies to comments

12:01 < eprodrom> I need to leave; thanks all!

12:02 < tantek> good to hear you eprodrom !

12:02 < cwebber2> ack ajordan

12:02 * Zakim sees no one on the speaker queue

12:02 < cwebber2> ajordan: go ahead

12:02 < nightpool> sorry cwebber2 between the drop-outs and ramble i couldn't understand much.

12:02 < cwebber2> nightpool, I completely lost myself in the commentary

12:02 * cwebber2 I had something coherent and my brain fried when trying to articulate it

12:02 ajordan: I have two questions. tantek--just so i'm clear, what you've been experimenting with is POSSE into github, right? Not only federated git stuff, right?

12:03 tantek: I have one example so far of a p2p issue tracking interaction, because there are so few of us doing it now

12:03 < tantek>

12:03 < cwebber2> q?

12:03 * Zakim sees no one on the speaker queue

12:03 < Loqi> [Tantek Çelik] I think IndieAuth makes a lot of sense as the way to implement a federated login protocol to provide a "simpler UX for OAuth2" login for Gitea as this issue is named.

12:03 < Loqi> It would also be possible to provide support for the "with GitHub" portion (as ori...

12:03 < tantek>

12:03 < Loqi> [Aaron Parecki] What we really need is federated authentication, but that doesn't exist yet.

12:03 < Loqi> This sounds like a great use case for IndieAuth.

12:03 < Loqi> IndieAuth is an OAuth 2.0 extension, which avoids the centralized problems with existing OAuth soluti...

12:03 tantek: aaron p. POSSE'd a comment to a GitHub issue and then I replied to that from my own site, and that was sent p2p with Webmention instead of through github.

12:04 ... this is a protozoa of federated issue tracking.

12:04 ajordan: great, I think it's important that we remember indieweb people want to do this stuff as well, and make sure it's bridgable and stuff like that

12:04 tantek: I agree, I think that this is a space where we want to start experimenting

12:05 ajordan: I'm thinking about the question of how we model issue closing/ACLs stuff like that. I'm thinking that a lot of this would be made way easier if we didn't try to model all of that stuff

12:06 ... If the canonical server for a repository is server A, and server B has a way to say "would you reject this activity", that would give a lot more flexibility

12:06 ... but still allow the protocol to work

12:06 < yookoala> Thanks all!

12:07 cwebber2: I think that's it on time so let's close it out for this week. I'm going to reply to the thread (git-federation#5) that yookoala linked, and we'll see all of you in two weeks

12:07 tantek: And if any of you are in portland in 3 weeks, it would be great to see you at the indieweb summit, we'll probably have a hackday about this

12:07 < ajordan> nightpool++

12:07 < Loqi> nightpool has 12 karma

12:07 < cwebber2> =========== MEETING LOGGING ENDS =========