13:57:21 RRSAgent has joined #did 13:57:25 logging to https://www.w3.org/2026/05/20-did-irc 13:57:38 rrsagent, draft minutes 13:57:40 I have made the request to generate https://www.w3.org/2026/05/20-did-minutes.html Wip 13:57:44 rrsagent, make logs public 13:58:56 transcriber-bot, pause 13:58:56 scribe- 13:59:17 transcriber-bot resume 13:59:38 transcriber-bot start 14:01:01 transcriber-bot, start 14:01:09 JoeAndrieu has joined #did 14:01:18 swcurran has joined #did 14:01:22 present+ 14:01:24 transcriber-bot, resume 14:01:24 scribe+ 14:01:30 Will Abramson: Oh, there we go... 14:01:33 Otto Mora: Oh, resume, resume, resume, okay... 14:01:36 ... Yes 14:01:37 transcriber-bot, pause 14:01:37 Will Abramson: Instagram? Oh, there we go. Perfect. Alright, sound, thanks also... 14:01:37 scribe- 14:01:43 pdl-ASU has joined #did 14:01:49 present+ 14:01:59 present+ 14:04:22 transcriber-bot, resume 14:04:22 scribe+ 14:05:24 transcriber-bot, resume 14:05:24 scribe+ 14:05:37 transcriber-bot, pause 14:05:37 scribe- 14:05:40 transcriber-bot, resume 14:05:40 scribe+ 14:07:15 Otto Mora: Okay, yeah, I see it now... 14:07:18 ... That's good 14:07:22 Will Abramson: Do it now... 14:07:33 ... Yeah, okay. So it's working now, but what happens when you leave? 14:07:37 transcriber-bot, pause 14:07:37 ... Yes 14:07:37 scribe- 14:09:14 transcriber-bot, resume 14:09:14 scribe+ 14:09:27 Will Abramson: So, welcome everyone to today's special topic call. I think the agenda is pretty straightforward, we're just going to continue... 14:09:33 ... Where we left off last week, um, working through the different 14:09:37 ... did URL dereferencing examples 14:09:42 ... And we'll probably hand over to Stephen, but before I do that, welcome back 14:09:57 ... Manu, um, and I just wanted to ask, does anybody have any other things on the agenda? Manu, I was just saying we're just going to continue working through the examples of did URLD referencing. Uh, using Steven's… Slide deck 14:10:03 ... Um, I don't have anything else, I think if we've got nothing else from the group, let's just dive straight into that 14:10:07 ... Um… Steven, do you want to share your screen, and we'll 14:10:10 Stephen Curran: Sounds good... 14:10:12 Will Abramson: Great, thanks... 14:10:14 Stephen Curran: You can see my screen?... 14:10:17 Will Abramson: Yes... 14:10:21 Stephen Curran: Okay, so we did this one, um, where... 14:10:29 ... the referencing outcome is DidDoc, the strategy we call Didoc. We're coming up with a name for the strategies. Um 14:10:32 ... Enumeration of them, and 14:10:39 ... The client is expected to focus on the identified node in the DIDOC, identified by the fragment 14:10:44 ... So I think we all know that. Typically used for verification method, but can be used for any node in the 14:10:55 ... JSON LD, and then, um, MediaType plays into that, because if it's a JSON LD, then that would be right, but… Perhaps other problems 14:10:59 ... Then we got into this one, this is where we stopped, um 14:11:03 ... I proposed a 14:11:10 ... Solution… a potential solution to this, um, after the call last week, so 14:11:13 ... Um, in this link here 14:11:14 ... Um… what… what 14:11:18 Will Abramson: Steven, could you drop the Alex? Could you drop the slide deck in again?... 14:11:21 Stephen Curran: Yeah. Oh, sorry, yeah, absolutely... 14:11:25 Will Abramson: Um... 14:11:27 Stephen Curran: Um, in chat, or in... 14:11:29 Will Abramson: Yeah, if you can do it in the RC, that would be great... 14:11:30 https://docs.google.com/presentation/d/1pez4sEyJ11fawmbYKcyjlNQqII3cePbB/edit?usp=sharing&ouid=109116496535883458301&rtpof=true&sd=true 14:11:34 Stephen Curran: I could do both. Yeah. There's chat… There's IRC... 14:11:37 Will Abramson: Cheers. Thanks... 14:11:43 Stephen Curran: Okay, so, um, the idea I had, uh, for doing this... 14:11:50 ... Jump to it quickly, is basically 14:11:59 ... treat. Whoops, that didn't work. I've got the wrong link in there. Well done, Steven! Sorry about that 14:12:03 ... I can get it quickly, though 14:12:10 ... Okay, this was on, uh, Marcus… Marcus, or… yeah 14:12:13 ... Um 14:12:20 ... I think I may have an idea for the work of this. So basically, the thing that people don't like about 14:12:28 ... What we have today for service type is what gets returned. is a… either 14:12:38 ... a modified DIDDoc that includes only the identified services, so you basically edit the DIDDoc on the fly and remove 14:12:43 ... Everything that isn't referenced in the parameters 14:12:47 ... And you apply the relative rep if it's present 14:12:59 ... Or, if the, um, accept is, uh, of a list of URLs, you return an array of URLs, so a JSON array 14:13:12 ... of URLs, again, with that. I think nobody really likes this, modifying the DIDOC or providing the list of URLs. So what I'm proposing is that instead of that 14:13:19 ... Um, we treat this the same as… Um… this 14:13:25 ... Um, as if they had specified an ID 14:13:32 ... Um, but they have the added advantage, if they do this, that you… so you… you pass back the did doc 14:13:39 ... And the retrieval strategy did dock, but the… person, or the 14:13:44 ... Specification says, focus on the identified nodes based on the 14:13:53 ... Um, on the, uh, parameters, on the service or service and or service type parameters 14:13:59 ... And that way, A, you can identify multiple, and people can have multiple 14:14:03 ... But otherwise, you… you handle it, it's just 14:14:15 ... Um, potentially, if there's not multiple, it becomes the same as the previous one, which is, hey, you get a DIDOC back, and you're told to focus on. Um, the identified note 14:14:16 q+ 14:14:18 ... Does anyone think that would work? 14:14:24 q+ to say dereferencing needs to eventually get to a resource 14:14:25 Will Abramson: Uh, I wrote myself on the queue. Sorry, Manny, you can go first, Phil... 14:14:28 Manu Sporny: I… oops, sorry... 14:14:39 ... Sorry, I just noticed that I don't even have IRC up. I had a hard time following that, Steven, sorry. It was… Difficult for me to understand 14:14:42 ... what you were proposing, um, but I might 14:14:45 Stephen Curran: I mean, essentially what I'm proposing is... 14:14:50 ... Which… this amounts to… this 14:14:53 Phillip Long (Georgetown, T3, ASU: Where are you? Oh... 14:14:56 Stephen Curran: On the sharing screen... 14:15:02 ... Um, so you, you basically say, instead of the result of it 14:15:08 ... Is the same as if you had passed multiple fragments for the nodes that you are interested in 14:15:15 ... And you're telling the person, hey, here's the DID doc, and these are the nodes you should pay attention to 14:15:18 ... And do whatever they tell you to based on their type 14:15:19 ack Wip 14:15:26 Will Abramson: Um… yeah, I put myself on the queue to say, I think the problem with that is... 14:15:30 manu has joined #did 14:15:34 ... like, what's… what's the point in that? Like, it's not doing anything new apart from just doing a fragment in a different way 14:15:36 present+ 14:15:39 ... That's why I wonder if the distinction between service 14:15:42 ... As a query parameter, should be 14:15:42 present+ 14:15:49 ... we are retrieving the resource at this service, right? Like, it's 14:15:54 q+ to ask why we're talking about this if we don't want to support it? 14:15:56 ... explicitly about fetching the result, and then if you want to access a node in a DID document, you should use fragment 14:16:05 ... There's no point, like, overloading that capability. I guess, apart from your saying, like, service type maybe allows you to indicate multiple nodes in the document 14:16:16 Stephen Curran: Yeah, so the two things is, um, it does allow you to have multiples, which in some use cases is interesting. Um... 14:16:21 q? 14:16:23 ack JoeAndrieu 14:16:24 JoeAndrieu, you wanted to say dereferencing needs to eventually get to a resource 14:16:25 ... Oh, shoot, I had a second. More relevant point, I apologize, so… Cool. Yeah. Cool 14:16:28 Will Abramson: Oh, we have a queue. Okay. Uh, Joe?... 14:16:35 Joe Andrieu: Yeah, your last point first, Steven, because it's a fragment, uh, it can't support multiples... 14:16:38 ... So, the problem with service type is that you can… you can map 14:16:43 ... file service to the type, but IDs have to be unique, so when you make it to a fragment 14:16:50 ... then did Document can only have one file service. Um, so that's… that's one wrinkle. Um 14:16:53 Stephen Curran: No, I'm saying that's a benefit, I'm saying that's a benefit, because instead of simply... 14:16:56 Joe Andrieu: No, no, you just said it could work with multiple, and I'm correcting you, it cannot work with multiples, because... 14:16:57 Stephen Curran: Oh... 14:17:00 Joe Andrieu: Fragments need to point to unique IDs... 14:17:09 Stephen Curran: Joe, you're not… you're not listening to me. Um, I didn't say you would use a fragment, I said it would be the equivalent of having... 14:17:14 ... Logically, multiple fragments, which is illegal, but allow the 14:17:15 q? 14:17:24 ... Um, the person that is dereferencing to realize, hey, here's the DIDDoc, and there's actually 3 nodes 14:17:35 ... in it that are relevant, and they should do what they should do based on the types of those 3 no's they get back. That's what I'm saying. I agree that you can't do it with a fragment 14:17:41 Joe Andrieu: Okay, so I have some other comments. One is, I do think we need to follow... 14:17:47 ... Um, for all of these did URLs, we need to follow the dereferencing to a resource 14:17:52 ... Until we have the resource that is in the local context, we have not completed dereferencing 14:18:03 ... And so, just returning a URL of any kind, I think, does not satisfy that. And that goes back to… so the place we got last week was this disagreement here 14:18:05 dmitriz has joined #did 14:18:10 ... where I had said this is a resource, Marcus says it's a filtered did doc. You're proceeding as if it's a filtered didc, Steven 14:18:20 ... And we might have that in an intermediate state like that, because I do think that the client is going to have to process that did document, and if what the client is looking for is any 14:18:25 ... service of a particular kind, then they're gonna use typical JSON parsing, and they're gonna figure that out 14:18:31 q+ 14:18:38 ... But then they still need to get to a resource, and it's our properties, um, and sort of the complex overlay of different layers at which you could specify how to get a resource. Which is why we need to sort of explain that 14:18:38 ack manu 14:18:38 manu, you wanted to ask why we're talking about this if we don't want to support it? 14:18:43 Will Abramson: Okay, thanks, Joe. Uh, my name?... 14:18:54 Manu Sporny: Um, I, I, I agree with, uh, what Joe said, and I think that might be an easier way to simplify... 14:19:00 ... Kind of the discussion, meaning, like, hey, we're really… we need to finish the dereferencing process to get to 14:19:18 ... you know, a thing. I guess I want to just not have this discussion at all, because I think it's eating up a lot of our time. Like, I don't think we need… we don't have a use case for service type. Let's take it out of there 14:19:24 +1 for removal 14:19:24 ... Right? Like, can we… can we talk about that? Because if we talk about that, then a lot of these slides become no ops 14:19:35 +1 for removal 14:19:35 ... Like, we're just not gonna talk about it, because, like, we're gonna move it into an external specification. People can go over there and talk about the external specification, the extension, how it's processed there 14:19:37 Stephen Curran: Okay... 14:19:40 Manu Sporny: And we cannot use valuable, you know... 14:19:40 Will Abramson: Mm-hmm... 14:19:43 Manu Sporny: Working group time to have all those discussions... 14:19:48 ack swcurran 14:19:50 Will Abramson: Yeah, thanks. I agree, we should… we already agreed we'd move service type into a, um... 14:19:52 ... extension. Steven? 14:19:55 Stephen Curran: Do we want to talk about what we do with service?... 14:19:58 Will Abramson: I think we should... 14:20:00 q- 14:20:00 q+ 14:20:01 Stephen Curran: Move on... 14:20:02 ack manu 14:20:04 Will Abramson: Come on... 14:20:07 Manu Sporny: I… well, QQQQ, Q. Um... 14:20:17 q+ 14:20:21 ... I, uh… I think if, Steven, if you feel, or anyone else on the call feels that service is special, and we should talk about it, we should talk about it, but I feel like, to me, like, I don't have a 14:20:28 ... deep, you know, understanding of all the corner cases with service, but to me, service kind of falls into the same category as service type. It's kind of like, who's using it? 14:20:30 q+ to talk about service 14:20:31 Stephen Curran: Yeah... 14:20:41 q- later 14:20:47 Manu Sporny: like, why can't you just use a fragment identifier for the 80% of cases? You know, like, do we need to talk about it? So unless someone's like, yes, we absolutely need to talk about it, I think it falls in the same category as service type. And I think every parameter, by default. He's just moved out, um... 14:20:50 ack JoeAndrieu 14:20:50 JoeAndrieu, you wanted to talk about service 14:20:54 Will Abramson: Okay, thanks. Uh, Joe?... 14:21:09 Joe Andrieu: Yeah, I think there's an important use case for service, um, and I do think we need to talk about it. One of the challenges with service right now is it, um, entangles relative ref, which is another thing that is, um... 14:21:14 ... is broken. It does not perform the way that I think we thought it did when we put it in the spec 14:21:24 ... Um, that difference, Manu, is to actually get back the resource. So, I want to put a did URL and a source attribute on an image tag 14:21:31 ... in an HTML page. And I don't want to get back… I don't want the browser to try and look at a DID document and say, oh, I'm supposed to represent this 14:21:35 Stephen Curran: Mm-hmm... 14:21:38 Joe Andrieu: I want it to get all the way to a resource that can go through linked resources, did linked resource, path service, whatever it is... 14:21:51 ... and get back the actual PNG so that the browser can handle it. And service, to me, has always said, hey, there's a service endpoint in here, you're going to dereference that service endpoint, and that's what this. DID URL actually points to 14:21:53 ack Wip 14:21:53 q+ 14:22:01 Will Abramson: Yeah, uh, I mean, I was just going to say pretty much that. I think that is the distinction, right? Service is... 14:22:07 ... execute the referencing for this, like, dereference this URL, and fragment is 14:22:13 ... you're accessing this node in the JSON-LD document, which, you know, maybe you might do something else with it 14:22:16 ack manu 14:22:21 ... But they are different things that you would expect to get back from the DRocking scale. I think that is a useful distinction. Uh, man 14:22:25 Manu Sporny: Is it possible for us to externalize that from the spec?... 14:22:32 ... Like, like, if there's an extension spec and there's a certain, you know, there's a service extension 14:22:40 ... can we externalize it? It feels like we could… I understand, like, we… it would be nice to have it in the spec, but, like, if it 14:22:46 ... It feels like a good test on whether or not externalizing these parameters 14:22:52 ... Uh, can fit into the generalized dereferencing algorithm in the data resolution spec 14:22:56 q+ to say "no. of course not" 14:22:58 ... My hope is that we can say, yes, of course, because that means that if we can 14:23:03 ack JoeAndrieu 14:23:03 JoeAndrieu, you wanted to say "no. of course not" 14:23:04 ... You know, move service into an extension spec, then we can move. Just about any parameter into an extension spec 14:23:08 Will Abramson: Uh, Joe?... 14:23:18 Joe Andrieu: Yeah, unfortunately, I would go the other way, man, I would say, no, of course not. The… this ability for the URL to dereference to something that's not the did document... 14:23:24 ... That's the whole point of the service endpoint, and I think it's a key part of how DIDs are meant to be used 14:23:28 ... And I think if we put that into a service, into a 14:23:33 q+ to say: I'm fine w/ that. 14:23:39 ... And to an extension, then we are just exacerbating the possibility of many, many, many different ways that people might, uh, do retrieval and try and represent resources 14:23:46 +1 to Joe 14:23:49 ... I think our work is to figure out how do we reconcile what Stephen wants to do, and what did Linked Resources do, and what Chuck did with did Linked Resources. Like, the reconciliation of that 14:23:55 ack manu 14:23:55 manu, you wanted to say: I'm fine w/ that. 14:23:57 ... Um, is what's gonna make DIDS useful as actual URLs and actual web content, and I think we need to push for that 14:24:01 Will Abramson: No... 14:24:09 q+ 14:24:10 Manu Sporny: Yep, yeah, and I'm fine with that. I mean, it looks like Steven and Joe agree that we can't push this out, or we shouldn't, um, and this is kind of the core... 14:24:14 Will Abramson: Okay... 14:24:16 ack Wip 14:24:17 Joe Andrieu: I was just laughing, it checked, being Chuck... 14:24:29 Will Abramson: Okay, I got myself a huge thing. I want to know, like, are we all agree that then service, like, if this is saying service equals 5 file service... 14:24:35 ... then… well, it'd really be the identifier of a service in the document, right? Then the 14:24:43 ... The approach would be, okay, what type is that service? And then execute the retrieval strategy of that type to retrieve the resource 14:24:44 Stephen Curran: Please... 14:24:47 Will Abramson: Because that is… that is different to the current... 14:24:52 ... SpecText, right? Current SpecText is expecting to just get back a service endpoint URL 14:24:59 ... Um 14:25:03 ... Yeah 14:25:09 Joe Andrieu: Yes, the type determines the strategy. I can imagine a type that actually supports multiple strategies. Um... 14:25:12 q+ 14:25:14 Stephen Curran: Okay... 14:25:16 ack manu 14:25:19 Will Abramson: What? Manny... 14:25:26 Manu Sporny: Um, is it clear what spec text needs to be written here, Joe, Steven, or is this the thing we're trying to figure out?... 14:25:28 Stephen Curran: I think we'll try to figure that out... 14:25:31 Manu Sporny: Okay... 14:25:34 Stephen Curran: I think we're still too early, because there's a lot more variations to come. And in particular... 14:25:46 ... This is the standalone version of where we have this, but what happens when it's combined with a pack? What happens if it's combined with a fragment? So we still need to keep going 14:25:55 q+ 14:25:58 ... Um, I wrote down here type of the identified service, and then I wrote object, because that's come up before 14:26:02 ... Um, if the query parameter 14:26:08 ... I mean, we're looking at particularly the service parameter, but what if it's some other extension? 14:26:10 ack Wip 14:26:11 ... Parameter that identifies something else 14:26:24 Will Abramson: I put myself on the computer said that I think that is just something that we would… is out of scope almost. Like, this parameter is saying, specifically, I am targeting a service... 14:26:31 ... by an identifier in the deduction. It's not talking about other kinds of objects that might or might not be in the deduction 14:26:34 Stephen Curran: Yeah... 14:26:38 Will Abramson: I agree, there might be other query parameters that people define in extensions, like the linked resource parameter that would work in a similar way, but… I think we shouldn't... 14:26:38 Stephen Curran: Okay... 14:26:41 Will Abramson: For you to travel... 14:26:44 ... Yeah, let's move on. Good 14:26:54 Stephen Curran: Okay, so here's one where we've got the combination. Um, I will take out type for this example and say it's service... 14:26:57 ... Um 14:27:05 q+ 14:27:07 ... To me, this becomes either invalid or valid only if the two things happen to line up. What do others think? 14:27:12 Manu Sporny: What do you mean by, uh, if the two things line up?... 14:27:14 Stephen Curran: Well, actually, no. Actually, I think... 14:27:17 Manu Sporny: Meaning you get the resource, and the resource has a fragment in it that matches?... 14:27:24 Stephen Curran: There's a couple of things, yeah. Um… What does the fragment... 14:27:29 ... talk about… does it talk about the DID doc, or does it talk about a resource 14:27:31 q+ 14:27:36 ... Uh… a resource, um, that you retrieve 14:27:39 ... After you've pulled in the DID doc 14:27:46 ... Uh, and then processed it, identified a resource, retrieved the resource, and then 14:27:47 ... the fragment apply for that, I'm not sure. Those are options, I see 14:27:50 ack JoeAndrieu 14:27:52 Will Abramson: Mm-hmm. Uh, I see... 14:27:56 ... I see Joe on the queue, let's do the queue. Joe? 14:28:04 Joe Andrieu: Yeah, I think the normal URL processing rules work here, which is... 14:28:11 ... Right, typically in a browser, right, the fragment doesn't go to the server, the server might have a redirect 14:28:17 ... And then, no matter what the number of redirects are, whatever the final resource is returned 14:28:18 q+ to agree, fragment is on final resource 14:28:23 ... the client applies that fragment to it. So I think we should proceed in the same way that we're gonna 14:28:28 ... treat the URL without the fragment, get that resource, um 14:28:33 ... Including whatever path or retrieval strategy we need to do 14:28:38 ack dmitriz 14:28:38 ... And then you interpret that fragment according to the media type of that resource 14:28:43 Will Abramson: Mm-hmm. Thanks, Joe. Uh, Dimitri?... 14:28:51 Dmitri Zagidulin: I think part of the challenge with this discussion is that... 14:29:00 q+ 14:29:01 ... what's on the screen, just a service and a fragment, doesn't mean anything. That is a meaningless set of parameters, it's not a valid thing to set. To send to a resolver 14:29:10 ... like, it's… it's meaningful syntactically, but… as in, that's a valid URL syntax, but it 14:29:14 ... It doesn't mean anything as far as the resolver, like 14:29:16 q+ 14:29:17 ack manu 14:29:18 manu, you wanted to agree, fragment is on final resource 14:29:21 Will Abramson: Okay, thanks. Manu?... 14:29:33 Manu Sporny: I think, Dimitri, the… at least my interpretation of this is, what would you do in the dereferencing process, which involves a resolver?... 14:29:37 ... Uh, if you saw something like this. Uh, and I agree with Joe that 14:29:49 ... you know, you… we want to follow what happens with URLs, which is, like, you are going to get the resource, and that fragment identifier applies to the final resource retrieved 14:29:49 ack JoeAndrieu 14:29:53 ... That feels like a fairly consistent way to interpret it 14:29:56 q+ JoeAndrieu 14:29:57 Dmitri Zagidulin: I agree with you, and Joe's right in that algorithm... 14:30:01 ... Uh, my point is that, by itself, the service 14:30:09 ... parameter is meaningless without a relative ref. Like, those parameters were meant to be used as a pair 14:30:15 ... By itself, it's just… we're just doing a combinatorial combination, combinatorial explosion 14:30:21 ... Of ways to parse JSON, of accessing a property 14:30:23 Will Abramson: Hmm... 14:30:26 ack Wip 14:30:28 Dmitri Zagidulin: And Jason, why are we doing that? Like, what is… what is that saving. Uh, clients... 14:30:32 q+ 14:30:34 Will Abramson: Okay, thanks, Dimitri. So I put myself on the queue to kind of agree with Joe and Manu, like... 14:30:39 ... I think this works today in the current spec text, and it works in 14:30:45 ... like Joe's algorithm. And I think the difference is, Dimitri, is if you use service like this 14:30:55 ... Uh, as we just discussed on the previous slide, and that Cirque service parameter is about retrieving the resource. Identified by this service 14:31:01 ... Then that is different from just a combinatorial. of, like, accessing JSON properties 14:31:08 Dmitri Zagidulin: But there's no… hold on a second. No, no, that's not valid, though, because we're not… we're not passing along a path to the resource... 14:31:11 Will Abramson: Good, that's it... 14:31:12 ack JoeAndrieu 14:31:15 Joe Andrieu: Sure, there is. There's… there's a queue... 14:31:18 Will Abramson: Uh, well, that's the cute, Joe, you can go. Joe?... 14:31:28 Joe Andrieu: Yeah, so that service defines an endpoint. That endpoint is pointing to a resource. So whatever that endpoint is, is the URL that you're going to get a resource from... 14:31:39 ... And then you see if it has a way to interpret the key one fragment. Um… No, I'm 14:31:44 ... I am hearing one thing that you're saying that relative ref and service were initially created together 14:31:52 ... Um, but relative ref is its own problem. So, if the reason you don't like this is because you like relative ref, which 14:31:58 q+ 14:32:00 ... probably… we might want to defer that and have the conversation about relative ref and what the problems are there, because I think if you see them bound together, we should probably 14:32:02 q+ 14:32:08 ... You know, figure out the solution together, which might be, frankly, Stephen Curran's, you know, attempt to get the path part. Uh, to do some of the lifting here 14:32:08 ack swcurran 14:32:12 Will Abramson: Uh, student?... 14:32:15 Stephen Curran: Um... 14:32:21 ... Yeah, I think this is a really fundamental thing, which is we're saying that 14:32:33 ... Will you referenced the current spec. The current spec is not even close to what we're talking about here. It says, return the alter did document or a list of URLs. I mean, that 14:32:39 ... to me is not close to what Joe was saying, which he's saying, oh, you've got a service 14:32:48 ... Retrieve what is implied by that service's endpoint, the service endpoint attribute. return it 14:32:51 ... Along with… and then the, um 14:33:02 ... apply the fragment after that gets returned. That's quite different from returning the altered did doc or a… Or a list of URLs. Um 14:33:08 ... So, I think if we're saying this, that they did resolution spec 14:33:12 ... Goes as far as saying, hey, when you have a service 14:33:17 q- 14:33:20 ... Um, the meaning of a service is you're supposed to resolve the, uh, in some way, resolve the endpoint 14:33:24 ... That's, uh 14:33:27 ack Wip 14:33:29 ... that's quite a fundamental change, and I think a little different from what I've heard previously 14:33:33 Will Abramson: Hmm. Yeah, it's like... 14:33:37 Stephen Curran: I'm not against it, because I kind of like it, but it's different from, you know... 14:33:45 ... What I've heard, I think, Manu say, which is, oh, the spec stuff that you don't actually go retrieve the resource 14:33:49 ... This is saying, we retrieve the resource, and then we apply the fragment 14:34:03 Will Abramson: Yeah, totally, and I just want to say, I mean, I do agree the spec text is different, but what is the same is, first, we apply the, like, there's two algorithms currently defined. The first algorithm really is talking about everything independent of the fragment, and then you put the fragment on at the end... 14:34:06 Stephen Curran: Yes... 14:34:10 Will Abramson: And I think that is the same… that is what we're talking about here, what Joe said, right? Like, you… you do this... 14:34:14 ... I mean, in his algorithm, he's called this thing, like, a base URL 14:34:23 ... basted URL, right? We do all the stuff, and then we apply the fragment at the end to the final resource that comes from dereferencing that basedid URL 14:34:26 q? 14:34:27 ... I think this makes sense. I feel like we should move on to get to the more complicated bits 14:34:30 Stephen Curran: Okay... 14:34:34 Will Abramson: You can always revisit these as we disagree... 14:34:40 Stephen Curran: Okay, so this one, we have just a plain path. Nothing else, just a path... 14:34:44 ... Um 14:34:54 ... That is what I hope, um, the algorithm that I put in, and I still think it's, it's pretty close to being right, is 14:34:57 ... Intended to do, which is, you just have a path 14:35:04 q+ 14:35:05 ... um, slash credential slash degrees dot JSON, and… and you process it. Based on 14:35:12 ... Um, the did doc, in this case, just the did doc you have, because there's no metadata, but just the did doc you have 14:35:18 ... Um, and… That would lead to 14:35:23 ... Um, some way of figuring out which 14:35:29 ack Wip 14:35:30 ... Um… which object you're gonna use, and… process that 14:35:34 q+ to say there might be metadata 14:35:36 ... So I think it's… and… Yeah 14:35:42 Will Abramson: Mm-hmm. I just put myself on the queue quickly to say, I think it's very hard to answer these questions, though, without the DID document. itself, right? Like, I think... 14:35:46 ... Maybe we don't need to see the DID document, but we need to… like, the DID document might have different things in it, and that's going to impact the 14:35:48 Stephen Curran: Yeah... 14:35:50 Will Abramson: The referencing outcome and the retrieval strategy... 14:35:52 ack JoeAndrieu 14:35:52 JoeAndrieu, you wanted to say there might be metadata 14:35:53 Stephen Curran: Absolutely. It totally will, yes... 14:35:57 Will Abramson: Uh, Joe?... 14:36:05 Joe Andrieu: Yeah, that was largely what I was gonna say. Also, there is metadata, so by this stage, we have performed resolution... 14:36:20 ... So we have… we do have the metadata back from resolution. So now, what happens here depends on what's in the did document and what's in the metadata 14:36:21 ... So, if there is a linked resource property, um, that has a path value of credentials 14:36:26 ... Then that will be, uh… the resource will be retrieved according to that 14:36:33 ... Um, I think if they have a path service in, um, then Steven would like his algorithm to apply 14:36:38 ... Um, and he's also advocated, right, that it could be a default 14:36:41 q+ 14:36:47 ... Um, and then did LinkedResources is gonna deal with it differently, um, if that path shows up in the metadata. Then, you've got to use their strategy 14:36:55 ... So, plus one to what Will said, we… the content of the did document affects what comes back here 14:36:56 ack swcurran 14:36:59 Will Abramson: soon... 14:37:04 Stephen Curran: Um, so, the reason there's no metadata is because that's another... 14:37:09 ... Example down further, so there is no metadata in this particular case that we're looking at 14:37:15 ... Um, I… I think I have an algorithm. The algorithm I've got 14:37:27 ... Um, I think deals with all of those cases. I wouldn't mind if I… Good. Um… Pop open a 14:37:32 ... Uh, different slide deck, but… okay 14:37:40 Joe Andrieu: Well, hold on, but before you do that, Steven, I just… I think we're probably in agreement on what goes in these fields, or at least there's some agreement here, and we… I just want to memorialize it before we get into your idea. I do want to hear your idea, though... 14:37:48 Stephen Curran: Okay, so I… I think in this case, it totally depends on what's in the div doc. The div doc, in this case, could have... 14:37:53 ... Um, pack service, or could have linked resources 14:38:02 ... Um, obviously did link resources doesn't come into play, but if we just add that to the, you know, we just go to that slide, we've got all 3 of them in play 14:38:10 ... And, um, I think the algorithm that I've put forward handles all 3 cases 14:38:13 q+ to suggest dereferencing outcome is a JSON file (assuming the server response happily) 14:38:17 ... And so I'd like to just make sure… I wouldn't mind if people will indulge me to say 14:38:23 ... To show you what the steps are. They're basically a similar type of four-step process 14:38:30 ... Um, that, uh, that is in Will's PR, but just for handling a path 14:38:31