2015-03-17 UTC
Guest, tantek, bblfish, tilgovi, Arnaud, Guest_, jaywink, elf-pavlik and peacekeeper joined the channel
SimonTennant, wilkie_, bblfish, the_frey, elf-pavlik, pfefferle and pete joined the channel
Augier joined the channel
Arnaud joined the channel
pfefferle joined the channel
cwebber2 joined the channel
# 12:36 tantek !tell Arnaud want to chair morning and I can chair afternoon? # 12:36 Loqi Ok, I'll tell them that when I see them next # 12:36 elf-pavlik i also just started drafting hypermedia API for adding media hosted on one domain to my online account hosted on another domain Arnaud joined the channel
# 12:47 Arnaud yes, not sure we will have a computer we can dedicate to that though # 12:51 Arnaud we have more time tomorrow so hopefully we can catch up then RRSAgent joined the channel
Zakim joined the channel
# 12:52 Zakim ok, trackbot; I see T&S_SWWG()8:00AM scheduled to start 52 minutes ago # 12:53 Zakim T&S_SWWG()8:00AM has not yet started, Arnaud # 12:53 Zakim On IRC I see RRSAgent, Arnaud, cwebber2, pfefferle, elf-pavlik, the_frey, wilkie_, SimonTennant, jaywink, tantek, danbri1, shepazu, KevinMarks, Tsyesika, sandro, Loqi, nickstenn, # 12:53 Zakim ... aaronpk, oshepherd, wseltzer, ElijahLynn, ben_thatmustbeme, mattl, JakeHart, bret, dwhly, bigbluehat, rhiaro, rektide, trackbot # 12:53 Zakim ok, Arnaud; that matches T&S_SWWG()8:00AM AdamB joined the channel
# 12:57 Zakim sees on the phone: confroom, elf-pavlik (muted) tilgovi joined the channel
# 13:01 AdamB Ann Bassetti (AnnB) - from Boeing, also chair social interest group # 13:02 AdamB Randal Leads - From organization Hypothesis, from the Annotations WG # 13:02 Zakim sees on the phone: confroom, elf-pavlik (muted) # 13:02 AdamB Benjamin Young - from hypothesis as well # 13:03 AdamB FJH - from annotations group # 13:03 AdamB ??? - from media goblin, implementing federation # 13:03 AdamB Chris Webber - from media goblin # 13:03 AdamB Matt Lee - creative commons, from gnu social projects # 13:04 AdamB Amy - phd student from ??? tilgovi_ joined the channel
# 13:04 AdamB Arnaud - chair of social wg, part of the IBM Open Standards group # 13:04 tilgovi <-- Randall Leeds (Hypothesis) # 13:04 AdamB Ben Roberts - from indie web group ben_thatmust_ joined the channel
# 13:05 ben_thatmust_ made it! fjh joined the channel
# 13:06 AdamB Arnaud: looking at the list we are still missing some people # 13:06 AdamB Zakim, who is on the phone? # 13:06 fjh s/ from annotations group/co-chair of annotation wg and device api wg/ # 13:07 RRSAgent I'm logging. I don't understand 'generate mintues', fjh. Try /msg RRSAgent help # 13:08 elf-pavlik no sound at all, also no welcome message asking to dial code after connecting via SIP AnnBassetti joined the channel
# 13:09 Zakim the conference code is 7625 (tel:+1.617.761.6200 sip:zakim@voip.w3.org), sandro # 13:11 AdamB we are working on technical difficulties in the room tantek joined the channel
# 13:13 AdamB Arnaud: talking about the agenda ...setup a straw man to get us started ... can get started with AS this morning # 13:13 AdamB ... tomorrow at the end left it open so we can fill it in with what we think would be good to do # 13:13 AdamB ... people have been adding to the agenda which is good # 13:13 AdamB ... there are no way we can address all the items that has been added to AS in the time allocated today # 13:14 AdamB ... we can discuss which ones are more important to talk about than others, maybe things that are more ready to discuss and can take advantage of the f2f meeting bblfish joined the channel
# 13:14 AdamB tantek: probably want james for the AS conversation rhiaro_ joined the channel
# 13:14 AdamB Arnaud: AS seems it is potentially our first victory # 13:15 sandro bblfish, can you hear on Zakim? Elf can't. We also are at talky.io/socialweb # 13:15 AdamB ... so in terms of w3c process, we have AS is a spec that has been published for a while and we need to focus on getting it to CR # 13:15 AdamB .. for the others its not as clear # 13:15 AdamB ... we need to take advantage of meeting to get far as possible jaywink joined the channel
# 13:16 AdamB ... the ceo has been putting more pressure on the working groups delivering on time fjh_ joined the channel
# 13:16 AdamB ... so as chairs we are responsible for keeping the working group on track so when time starts slipping by we need to consider cutting features down # 13:16 AdamB ... just like any other software shipping # 13:17 Zakim On IRC I see jaywink, rhiaro_, bblfish, tantek, AnnBassetti, fjh, tilgovi, AdamB, Zakim, RRSAgent, Arnaud, cwebber2, pfefferle, elf-pavlik, the_frey, wilkie_, SimonTennant, # 13:17 Zakim ... danbri1, shepazu, KevinMarks, Tsyesika, sandro, Loqi, nickstenn, aaronpk, oshepherd, wseltzer, ElijahLynn, ben_thatmustbeme, mattl, JakeHart, bret, dwhly, bigbluehat, rhiaro, # 13:17 AdamB ... so we need to keep this in mind so we can do what we can to ship # 13:17 AdamB ... it may not be what we want but it's still good # 13:17 aaronpk bblfish: zakim is still active so feel free to dial in and listen that way if you want # 13:18 AdamB ... we need to seriously think about that. so i would really like we have a clear understanding of what is left to AS to ship it # 13:18 fjh who is on the video, ? and henry? # 13:18 AdamB ... would be good do then do that for the social api as well harry joined the channel
# 13:18 tantek fjh it looks like elf-pavlik and bblfish are on talky # 13:19 AdamB ... we've been through a lot of work on it e.g. the user stories ... so we need to start narrowing down the things to work to get to a path that will get us somewhere # 13:19 AdamB ... have to bite the bullet and do something # 13:19 harry but since you are both on talky.io, let's try that first maybe? # 13:19 AdamB ... again, seems like AS is the lowest hanging fruit and if we can demonstrate some success will help with w3c management # 13:19 AdamB tantaek: want to underscore what Arnaud said, especially shipping and cutting feature # 13:20 AdamB s/tantaek/tantek/ # 13:20 AdamB ... we've tried to get things done but we've failed to make any date so far # 13:20 AdamB ... we need to ship something in order to keep our schedule that was committed too # 13:21 AdamB ... while we are trying to get AS to CR we should try to propose as minimal steps as possible to get social api and federation out the door # 13:21 bblfish anyway, nice to see people's face even if very far away # 13:22 Zakim sees harry, AnnBassetti on the speaker queue # 13:22 Loqi AnnBassetti: rhiaro left you a message on 1/15 at 2:03pm: Try pandoc.amy.gy - improvements/bug reports welcome # 13:22 AdamB fjh: i think what you are saying makes sense # 13:22 AdamB ... i'm assuming you could cycle vocabularies pretty quickly # 13:22 AnnBassetti loves the loqi feature that tells people stuff # 13:22 AdamB ... annotation as an action # 13:22 elf-pavlik I understood, we can update Vocabulary much easier than change Core # 13:23 AdamB tantek: there are active impl of AS # 13:23 AdamB .... some are previous version of is, not clear how quickly they may move to the next version # 13:23 AdamB ... at exiting CR things have to be marked as implemented # 13:24 AdamB ... we haven't labeled anything out of scope yet eprodrom joined the channel
# 13:24 aaronpk there's some neat experimental electronic music happening in bblfish's room # 13:25 bblfish trying to understand what channel one is hearing better on # 13:25 Zakim tantek, you typed too many words without commas; I suspect you forgot to start with 'to ...' # 13:25 Zakim sees harry, AnnBassetti, tantek on the speaker queue # 13:25 AdamB Harry: i would hold out to CR phase which would make some people really happy, but to me the goal of the wg is to maximize the conferment impl # 13:25 Zakim sees harry, AnnBassetti, tantek, eprodrom on the speaker queue # 13:25 Zakim sees AnnBassetti, tantek, eprodrom on the speaker queue # 13:26 AdamB ... so i would say don't worry about testing right now # 13:26 AdamB ... we could always re-edit and the w3c process is changing to simplify things # 13:26 AdamB ... it is changing so you can go back and edit things # 13:26 Zakim sees AnnBassetti, tantek, eprodrom on the speaker queue # 13:26 Zakim sees tantek, eprodrom on the speaker queue # 13:26 AdamB AnnBassetti: in terms of choosing things to happen, is the intention to vote early and often # 13:27 AdamB ... to establish an early set of things to do ? # 13:27 AdamB tantek: that is one method we can chose to get things out the door, and i am in favor of doing that # 13:27 Zakim tantek, you wanted to clarify entering vs. exiting CR # 13:27 Zakim sees eprodrom, bblfish on the speaker queue # 13:27 AdamB AnnBassetti: yeah thats what i'm asking, is that what the group would like to do? # 13:27 fjh s/saying makes sense/saying makes sense, question on whether we can still add to vocabulary, for example adding annotation action/ # 13:27 harry I'd rather rush to see implementation experience :) # 13:28 AdamB tantek: harry you mentioned rushing to CR. there is a diff between entering CR and exiting CR. we can enter whenever we want and that is when we think the spec is detail enough # 13:28 AnnBassetti my fear is, if we try to do it all, at the beginning, we'll never get there # 13:28 AdamB ... exiting CR is different, where we as a group have more flexibility where we will have to make more judgement calls for the group # 13:28 AdamB ... we can give it more CR time for tests etc # 13:28 harry However, you can implement out of Working Drafts, that's common in many Working Groups - and we can start getting interop in Working Drafts, that's what most successful WGs do nowdays. # 13:28 AnnBassetti seems like it's better to commit to some initial stuff, get it done, then quickly jump on the next set # 13:29 AdamB ... the risk of rushing it through we might have to drop features in order to get to exit # 13:29 AdamB ... i'm in favor of an aggressive CR schedule # 13:29 harry when success = number of interopable implementations (ideally with real users , real active open source implementations, and real products) # 13:29 harry To be precise, once you exit CR you can't really change the spec easily, although we are working on changing that in terms of W3C process. SimonTennant joined the channel
# 13:30 AdamB eprodom: my question is somewhat about the process, in terms of data serialization what are we looking for? parsers? # 13:30 AdamB ... are there meaning test that we do or is it we say you can do this with a json-ld parser, ok then done # 13:30 AdamB sandro: think we base it off the community # 13:31 AdamB tantek: there is some experience in w3c in this. references xml history # 13:31 AdamB .... there are lot of xml parsers out there so obvious these are compliant but turned out that wasn't true # 13:32 AdamB ... we should not play that approach in this group. just because its based on an underlying technology doesn't mean it's compliant to the spec # 13:32 AdamB ... implementors get to a point and say, well how do i know that i'm compliant to section 2.4.6 ? # 13:32 AdamB ... needs to be a test to prove it out # 13:33 AdamB eprodom: so would a way for us to satisfy this, the testing and implementation concepts, is a way for us to move forward this is n many documents and these are valid and these ones are not and your implemetnation can parse them and figure that out? # 13:34 AdamB ... mock AS documents to drive the test? # 13:34 AdamB sandro: probably need more, may not be enough # 13:34 AdamB tantek: if you can't figure out a test to figure out the results of two different features then you must not need them both # 13:34 elf-pavlik q+ re: we need more AS2.0 examples and we could use them in our tests # 13:34 Zakim sees bblfish, harry, elf-pavlik on the speaker queue # 13:35 Zakim sees bblfish, harry, elf-pavlik on the speaker queue # 13:35 AdamB Arnaud: guess we can say we have started the conversation about AS # 13:36 AdamB Arnaud: first evan, we have to figure that out, there are no hard and fast rules in the w3c # 13:36 AdamB ... the w3c has moved to having test suite etc # 13:36 AdamB ... one thing that is not up to us is the criteria to actually enter CR. there used to be a last call step which happened when all your issues have been closed. # 13:36 Zakim sees bblfish, harry, elf-pavlik, fjh on the speaker queue # 13:36 Zakim sees bblfish, elf-pavlik, fjh on the speaker queue # 13:37 AdamB ... we are telling the world we think we are done # 13:37 AdamB ... then had to go through the disposition of the comments from the public, then after that you could go to CR. # 13:37 harry my notes about links to the 'current state of testing' is in IRC, and has been enabled to the list # 13:37 AdamB ... the w3c has gotten rid of the last call step # 13:37 AdamB ... we just did this with the LDP group # 13:38 AdamB ... have to show the "director" yes we reached out to the world and published a spec that was stable and we've properly responded to comments AnnB joined the channel
# 13:38 AdamB ... its good to have a test suite which is kind of expected now and typically have a link to the test suite from the draft so that people can find it easily # 13:38 harry We now have a nice github process for testing, although many WGs don't know about it. # 13:39 AdamB ... doesn't have to be final so that people can see it and "stuff" # 13:39 AdamB ... and have some frameworks that can collect the results of the test # 13:39 harry It will at least be relevant to the API, although some of the more behavioral testing may be tricky as its optimized for browser APIs # 13:39 tantek test -> implementation report per implementation -> see which features have 2+ implementations that pass all tests for that feature # 13:39 AdamB ... we can raise the bar and ask the full implementation if we wanted # 13:39 Zakim sees bblfish, elf-pavlik, fjh, eprodrom on the speaker queue # 13:40 AdamB ... we need to meet all of that so for the exit we can go to w3c and have this proof # 13:40 harry sounds like we are now in ActivityStreams 2.0 discusion, so let's change topic as soon as the queue is closed. # 13:40 AnnB waves at bblfish and elf # 13:40 AdamB ... in the w3c process there is a feature called 'feature at risk' jasnell joined the channel
# 13:41 AdamB ... we can mark some that way and if we are not meeting them we can simply remove them and procede jasnell_ joined the channel
# 13:41 AdamB ... if we don't do that we have to go back and republish a spec in CR again # 13:41 tantek q+ to say why everything is at risk by default: absent evidence (test + implementation report), we don't know if anything is implemented. otherwise we have to do extra work of cutting later with CR cycle. # 13:41 Zakim sees bblfish, elf-pavlik, fjh, eprodrom, tantek on the speaker queue # 13:41 AdamB ... anything that changes conformance you have to republish # 13:42 Zakim sees elf-pavlik, fjh, eprodrom, tantek on the speaker queue # 13:42 AdamB ... it is a great way to provide some safety valves in the spec # 13:42 AdamB henry: thank for that review, it was a good one. # 13:42 AdamB ... on the implementations I think we need to make sure we have the archticture right so it can grow as the space grows # 13:43 AdamB ... at that point want to start with really basic thing because if we do it right can later do more complicated things # 13:43 Zakim elf-pavlik, you wanted to discuss we need more AS2.0 examples and we could use them in our tests # 13:43 Zakim sees fjh, eprodrom, tantek on the speaker queue # 13:43 Zakim sees fjh, eprodrom, tantek on the speaker queue # 13:43 AdamB elf: for testing i think we need come up with examples # 13:44 AdamB ... and also use test to verify that the examples show usage of the spec # 13:44 AdamB ... to verify the expected functionality # 13:44 AdamB eprodrom: agrees and there are some great examples in the stories # 13:44 AdamB ... maybe something along, here is an example and show me the actor in that example # 13:44 Zakim sees eprodrom, tantek on the speaker queue # 13:45 AdamB ... but we do have a large set of examples to start with but not a lot of counter examples # 13:45 AdamB fjh: thinking about what tantek said, i'm wondering what the implications of that # 13:45 AdamB ... so under object have all of these properties # 13:46 AdamB ... assuming other apps will come up with new types of things you can anticipate everything # 13:46 harry If a vocabulary item isn't used, I support removing it from spec # 13:46 Zakim sees eprodrom, tantek, bblfish on the speaker queue # 13:46 harry We can always put more experimental vocabulary stuff in an IG "Extended Vocabulary" note. # 13:46 AdamB tantek: fundamental point is that if there is a feature in the spec then we as a group believe that implementors must implement it # 13:46 AdamB ... unless it is marked optional # 13:46 Zakim sees eprodrom, tantek, bblfish, cwebber on the speaker queue # 13:46 AdamB ... if so we are obligated to provide a test # 13:47 harry However, this does not mean every implementation should test *every* feature, but that each feature should have at least two implementations # 13:47 AdamB fjh: let me be concrete in a diff way, say for example the post action, do you need a full blog platform to do that # 13:47 jasnell_ quick note: the spec doesn't say anything about what these various objects SHOULD *do*. It just says what those things *are*. There are no conformance requirements beyond syntax # 13:47 AdamB tantek: so it depends on what the spec says # 13:47 Zakim sees eprodrom, tantek, bblfish, cwebber on the speaker queue # 13:47 AdamB fjh: so for annotations we'd have to be very careful # 13:48 AdamB tantek: if you can't verify the feature you are not done with the spec # 13:48 AdamB Arnaud: we have to be careful when it comes to vocabulary # 13:48 trackbot issue-16 -- better separate grammar/vocabulary and improved spec structure -- open # 13:48 Zakim sees tantek, bblfish, cwebber on the speaker queue # 13:48 AdamB eprodrom: i think writing blog software is too far us initially # 13:49 AdamB .... my point of view somethign that could generate streams based on inputs # 13:50 AdamB ... if you know X generated some content at this time and you generated that based on the AS 2.0 spec. that seems like enough and shoudl test the data serialization etc without worrying about other things like stuff required for blogging softtware # 13:50 Zakim tantek, you wanted to say why everything is at risk by default: absent evidence (test + implementation report), we don't know if anything is implemented. otherwise we have to do # 13:50 Zakim sees bblfish, cwebber on the speaker queue # 13:50 Zakim ... extra work of cutting later with CR cycle. # 13:50 AdamB .. which is not important for us # 13:50 AdamB tantek: this is why i mentioned everything should be marked at risk by default # 13:51 harry I actually prefer to implement in Working Draft, but if people want to make these more painful in terms of rushing to CR, power to the people. # 13:51 harry The key is to implement early and get feedback to the Working Group # 13:51 AdamB ... if we don't mark feature as at risk we are expecting implementors to implement. but if we actually did that don't think it would be looked on as favorable # 13:51 harry and not put features you don't actually want and plan to implement in the spec # 13:51 AdamB ... if you want something to be required and not at risk it is your burden of proof for that # 13:52 AdamB ... but that is the challenge that should be required then ship it # 13:52 AdamB tantek: the sooner we have a test suite framework and the sooner people can start submitting tests # 13:52 trackbot issue-8 -- Test suite for activity streams 2.0 -- open # 13:52 Zakim sees bblfish, cwebber, harry on the speaker queue # 13:52 AdamB ... and issues are good candidate for tests # 13:52 tantek harry, I expect implementers to be attempting implementing ASAP to give feedback # 13:53 AdamB ... and can tell you from his experience that when WGs don't do that it holds things up # 13:53 AdamB ... so i really encourage that we do start on the test suite # 13:53 AdamB tantek: the test needs to prove the test but the expected result as well # 13:53 AdamB bblfish: this is really interesting stuff, test suites don't test everything # 13:53 trackbot issue-19 -- WG communication channel explosion -- raised # 13:54 tantek actually, the test suite IS expected to test every feature in the spec # 13:54 AdamB ... i propose we use issue-19 would be able to use this system to communicate within this group without using email # 13:54 AdamB eprodom: is that one of our goals for this WG that AS should be used to replace email? # 13:54 AdamB bblfish: i was just going little bit further # 13:54 harry That email use-case was missing from the user-stories I think # 13:55 trackbot action-45 -- Henry Story to Testing Activity Streams 2.0: explore ontology based testing -- due 2015-03-17 -- OPEN # 13:55 AdamB cwebber: very first part of this conversation was about AS test, what would those test look like # 13:56 AdamB ... AS don't really do anything until we start doing something with them ... like in the social api # 13:56 wseltzer rrsagent, pointer? # 13:56 AdamB ... what would the test be, is it just like changing an in memory db? # 13:56 Zakim sees harry, eprodrom on the speaker queue # 13:56 tantek q+ to answer cwebber2's question about how do we test something in AS # 13:56 Zakim sees eprodrom, tantek on the speaker queue # 13:56 AdamB arnaud: would like to prompt james to answer that # 13:57 AdamB harry: there are diff ways to test functionality .. some of teh tradiditional test suites in the w3c may not apply but there are things like selinium etc # 13:57 sandro q+ to answer Chris: you just leave that bit agnostic # 13:57 Zakim sees eprodrom, tantek, sandro on the speaker queue # 13:58 AdamB ... practical step might be to have a place to put tests # 13:58 Zakim sees eprodrom, tantek, sandro, elf-pavlik on the speaker queue # 13:58 AnnB s/student from ???/student from Univ Edinburgh/ # 13:58 Zakim sees tantek, sandro, elf-pavlik on the speaker queue # 13:59 harry I mean it's obvious just parsing JSON-LD is nonsense as a test # 13:59 jasnell_ AS2 currently is JUST a vocabulary. It only describes events and does not specify any kind of conformance beyond syntax. So the key question is: do we need to specify conformance requirements beyond syntax? # 13:59 jasnell_ this is the processing model question that Erik keeps raising # 13:59 harry The issue is that the output isn't uniform likely (i.e. it will change the state of some JS somewhere else) # 13:59 Zakim On the phone I see confroom, bblfish, jasnell # 14:00 cwebber2 tantek: I'd be happy to see that AS *doesn't* need the API to do tests # 14:00 tantek purely sloshing terms / syntax around is not really interesting, nor does it justify the diversity/complexity of the current specs # 14:00 cwebber2 tantek: I was just trying to figure out what it would look like without it :) # 14:00 Zakim sees tantek, sandro, elf-pavlik on the speaker queue AdamB_ and rhiaro joined the channel
# 14:00 harry I am personally having trouble imagining how we will test this as just a data format, but I imagine it might be possible. # 14:00 tantek cwebber2, there's a lot of history here with feeds publishing/consuming # 14:01 bblfish I think there are three level of tests: one for each spec, and one integration test suite # 14:01 harry I am against 'ontology'-based testing as the *only* tests although I'm happy for people to do that. # 14:01 harry I've never seen reasoners been used in real-world applications # 14:01 AdamB_ jasnell: up to this point the format has always been about describing the event and not specififying what you should do with it # 14:01 AdamB_ ... even back in AS 1.0 # 14:01 trackbot issue-12 -- Action Types Structure and Processing Model -- open # 14:01 AdamB_ ... i know this is a point that erik has raised a number of times # 14:02 harry I think if you go up to an ordinary web developer and say 'Hey, run this RDF-based reasoner to see if you are compliant' you are going to get blank stares # 14:02 AdamB_ ... in order for us to have meaning test, or even a like activity, we are going to have to define what you are to do with it # 14:02 AdamB_ ... in order for us to take that additional step with that # 14:02 Zakim sees tantek, sandro, elf-pavlik, eprodrom on the speaker queue # 14:02 harry I think for people that can run a reasoner, they should and that should be specified, but that should not be the only tests. # 14:02 eprodrom This is what the API should do # 14:02 AdamB_ ... but do we want to assign actual behaviors to those then we probably need to simplify # 14:03 harry noting that we have other things to figure out # 14:03 Zakim tantek, you wanted to answer cwebber2's question about how do we test something in AS # 14:03 Zakim sees sandro, elf-pavlik, eprodrom on the speaker queue # 14:03 eprodrom "When a user posts a LikeAction, the object goes into their 'liked things' collection, and the user goes into the collection of 'people who like this' for the object" # 14:03 harry ACTION: harry to set-up a github for AS2.0 testing (whatever that ends up being, we'll need a github!) # 14:03 trackbot Created ACTION-48 - Set-up a github for as2.0 testing (whatever that ends up being, we'll need a github!) [on Harry Halpin - due 2015-03-24]. # 14:03 AdamB_ tantek: james i think you identified some pretty core issues that i think we need to record, which is every key action in AS needs to provide ???? # 14:04 AdamB_ ... second issue is there is a lot in the spec and it's not clear we have implementor interest in doing something with everything in the spec so we need to gather that in some way # 14:04 AdamB_ .. is that a reasonable characteristic of the two concerns? # 14:04 elf-pavlik associated action-48 with issue-8 and AS2.0 product in a tracker # 14:05 AdamB_ jasnell: i documented uses case but we have no real data on actual usages of them # 14:06 jasnell_ The Atom Feed Validator (which focused on syntax only) was critical to ensuring interop among feeds # 14:06 Zakim sees sandro, elf-pavlik, eprodrom on the speaker queue # 14:06 AdamB_ ISSUE: need to get data on implementor interest on specific features of the AS spec # 14:06 harry +1 a validator, but I would like to make sure also that things 'happen' in terms of actual activities # 14:07 bblfish Mhh, there may be one simple test: that links link to resources that match the type that the link says they should have stevenroose joined the channel
# 14:07 Zakim sees sandro, elf-pavlik, eprodrom, bigbluehat on the speaker queue # 14:07 AdamB_ tantek: are the readers processing the information as we would expect # 14:08 cwebber2 I guess I don't have a concept of what "processing model" as a test here means :) # 14:08 AdamB_ ... if we need to put more language in the spec then lets do that # 14:08 harry are they "processing" it in terms of the 'processing model of the spec' and then 'does it do something' after the processing # 14:08 harry I think the tricky bit is the 'does it do something' # 14:08 Zakim sees sandro, elf-pavlik, eprodrom on the speaker queue # 14:08 AdamB_ cwebber and tantek discussing question about testing # 14:09 harry I think the RDF(S) reasoning is useful in terms of possible 'processing model' but we need to be explicit about this in the spec due to earlier 'blank stares' comment. An implementer should read the spec and write an implementation. # 14:09 Zakim sees sandro, elf-pavlik, eprodrom on the speaker queue # 14:09 AdamB_ cwebber: i'm not clear on what it means in the processing model # 14:09 AdamB_ ... what is it checking for # 14:09 AdamB_ tantek: it is good to start with a use case # 14:09 harry Without really knowing or caring about RDF. # 14:09 harry (although if they end up using RDF power, power to the people!) # 14:09 AdamB_ ... the spec needs to say if you consume one of "these" then you need to do "this" # 14:10 AdamB_ eprodom: the tough part of of AS 2.0 is there are 2 usages, one use is the logging format # 14:10 AdamB_ ... the other use is as a command language. where evan wants to tell his system 'evan posts a note' # 14:10 AdamB_ ... those are 2 different things # 14:10 AdamB_ tantek: it is great to specify both of those in the spec # 14:11 AdamB_ tantek: if you look at other specs like html, they have different classes of user agents # 14:11 AnnB we could hear your dog or something, bblfish # 14:11 AdamB_ cwebber: didn't realize until eprodrom mentioned the logging vs the command of AS # 14:11 harry when the kid grows up, he should have an AS2.0 compatible blog :) # 14:12 cwebber2 yes re: the logging format is one use, the other thing as a command language, I think the thing is that testing the command language is that it *does* things, and it look, that brings a lot of clarity to my confusion :) # 14:12 AdamB_ tantek: could be argued there is a little mini api in the AS spec # 14:12 AdamB_ ... here is how it works as a command api # 14:12 AdamB_ .. the social api goes far beyond the command language # 14:12 Zakim sees sandro, elf-pavlik, eprodrom on the speaker queue # 14:12 AdamB_ ... maybe the social api has a crud # 14:12 harry Yes the API might end up being pretty simple GET/POST (CRUD) with some "URL template" structure that maps to the vocabulary in some sensible manner. # 14:13 harry I think that's on the table for this afternoon. # 14:13 AdamB_ tantek: the key is to document the multiple possibilities .. we don't leave it open ended # 14:13 Zakim sandro, you wanted to answer Chris: you just leave that bit agnostic # 14:13 Zakim sees elf-pavlik, eprodrom on the speaker queue # 14:13 AdamB_ sandro: re: how do we test this without having that. the answer is we don't care about that, we are agnostic about that # 14:13 AdamB_ .... you can scaffold it together how ever you want # 14:14 Zakim sees elf-pavlik, eprodrom on the speaker queue # 14:14 Zakim elf-pavlik, you wanted to discuss issue-12 Action Types Structure and Processing Model # 14:14 bblfish well we can assume a minimal API already: HTTP no? # 14:14 AdamB_ elf-pavlik: would like to note issue-12 # 14:14 trackbot issue-12 -- Action Types Structure and Processing Model -- open # 14:14 trackbot issue-4 -- Do we rely on explicit typing or support implicit typing based on explicit property names? -- open # 14:14 AdamB_ ... would also like to bring up querying the data # 14:14 AdamB_ ... to query for AS data danbri joined the channel
# 14:15 AdamB_ eprodom: i would propose that for AS 2.0 we only test it in it's logging capacity. in understanding as what has already happened # 14:16 AdamB_ ... it might make some of the problem easier to test. can defer to the social api testing for the semantics of using AS as a command language or just leave it out entirely # 14:16 cwebber2 I hope I didn't drag that out, but clarifying this has been really helpful to me # 14:16 tantek +1 AS as feeds / consuming for tests first, command language later # 14:16 AdamB_ ... seems to me that the process of testing as command language is too much to bite off # 14:17 AdamB_ ... guessing low 100s for test cases. take the examples, parse out the properties etc # 14:17 AdamB_ tantek: it also makes it easy for the spec # 14:17 bblfish One could for example test things: a client follows links, and should get doing this what it expects to get # 14:17 cwebber2 so it seems that doing the activitystreams as logging, then transitioning towards activitystreams as a command language could be helping us moving towards the api test # 14:18 tantek btw anyone that thinks testing is "easy" is requested to provide URLs to tests :) # 14:18 AdamB_ Arnaud: lets try to close the topic for now. we have plenty of other things to cover # 14:18 AdamB_ ... the question is, who is going to take the lead on getting some sort of framework for getting test suite setup etc # 14:18 AdamB_ ... how do we make progress on this # 14:19 AdamB_ ... we need volunteers # 14:19 AdamB_ tantek: is anybody producing an AS accordance to the spec # 14:19 Zakim sees on the phone: confroom, bblfish (muted), jasnell # 14:19 harry So we need some open-source, open web implementations :) # 14:19 AdamB_ jasnell: they have one customer that is doing this but can't provide a public URL # 14:19 AdamB_ tantek: is anybody doing this on the web # 14:20 AdamB_ tantek: even seeing an example on the web # 14:20 AdamB_ ... would be a great start # 14:21 AdamB_ ... so before we jump ahead to 'how do we create a test framework' i say we get 1 example of the AS 2.0 spec # 14:21 AdamB_ jasnell: it parses the activity stream from github, its rudimentary # 14:22 trackbot Created ACTION-49 - Publish as2.0 data on one's own website [on Pavlik elf - due 2015-03-24]. # 14:22 AdamB_ Arnaud: let me ask who is actually planning to implement AS? # 14:23 AdamB_ ... so a lot of people are going to do it, so how are you guys going to test it ? # 14:24 jasnell_ fjh: it's a simple node app. it receives hook notifications from github and converts the events into AS2 objects stored in a cloudant db # 14:24 harry speaking of work, let's try to solve the rest of the AS2 issues # 14:24 tantek jasnell++ for providing a live AS2 stream for viewing! # 14:24 AnnB hopes AdamB will clarify what he's done with AS & how it # 14:24 aaronpk q+ to discuss my experience in "testing" microformats publishing/consuming # 14:24 Zakim sees elf-pavlik, aaronpk on the speaker queue # 14:24 AdamB bblfish: if you are publishing AS 2 and there is no client that is reading it out there, it would be extremely useful to have translators for previous versions # 14:25 harry GRDDL is a dead spec no one in their right mind should use # 14:25 Zakim elf-pavlik, you wanted to discuss publishing my data using schema.org or AS2.0 dilema # 14:25 harry it depends on @profile which was rightfully deprecated by HTML5 # 14:25 harry and depends on your web-page being valid XML # 14:25 AdamB elf-pavlik: i just wanted to mention that i started to publishing my information on my website and using AS # 14:25 bblfish the main point is that there is a lot of RSS out there and Atom # 14:25 Zakim sees aaronpk, eprodrom on the speaker queue # 14:25 harry However, yes, more power to shims to move older formats to AS2 # 14:26 bblfish so that means that any blog engine has to publish these other formats too # 14:26 harry would actually be useful to have an AS1->AS2 converter # 14:26 harry or from other widespread vocabularies - there are probably things in RDF space as well that could be converted over # 14:26 AdamB eprodom: i will volunteer to extract the examples from the main documents, pick properties and put together a test for those exmaples # 14:26 eprodrom testanything.org # 14:26 tantek bblfish: not at all - indieweb sites work peer to peer without any RSS or Atom # 14:26 AdamB ... probably use testanything to do so # 14:26 harry I also think a microformat->AS2 converter would be quite cool # 14:26 AdamB ... so could task js vs ruby library to test # 14:26 Zakim aaronpk, you wanted to discuss my experience in "testing" microformats publishing/consuming # 14:27 tantek it's up to individual sites, if you have a use-case for supporting legacy feed formats, you can do so on your site. it's just not required for interop # 14:27 harry Or if there are any RDFa/microdata compatibile sites, converting them over automagically to AS2.0 would be great # 14:27 AdamB aaronpk: was going to say back on the how do we test, problem i'm having with microformats ... until machines start consuming i don't know if it's corret # 14:27 AdamB s/corret/correct/ # 14:27 harry I don't think those would count as 'tests' per se, but would be excellent work to increase adoption and will end up being critical infrastructure # 14:27 bblfish well the html5 group did a lot of work on making sure that the extensions they did did not break existing web pages # 14:28 tantek q+ to also share microformats experience of big difference in "verifying" parsed output, and seeing what specific consuming uses do with it (showing comments, readers etc.) # 14:28 AdamB .. now that there are readers its been very helpful to have people looking at machine readable markup # 14:28 bblfish so harry are you going off the idea that we should ignore the past? # 14:28 Zakim tantek, you wanted to also share microformats experience of big difference in "verifying" parsed output, and seeing what specific consuming uses do with it (showing comments, # 14:28 harry bblfish, failed technologies should be forgotten. # 14:28 harry It's actually incompatible with existing web technologies and had no uptake # 14:28 AdamB tantek: i think the experience aaronpk is important. there is a huge difference in UI vs reader etc # 14:29 AdamB ... very different level of testing for finding flaws in specs etc # 14:29 cwebber2 so, Tsyesika and I are planning on doing a proof of implementation # 14:29 harry I had a good time learning W3C process by chairing that WG and met some great people, but in terms of a standard it was an epic fail. # 14:29 AdamB ... would caution that just verifying looking at the output will be enough. it is a good first step # 14:29 cwebber2 Arnaud: but the reason I'm not saying "we are doing this" is because I think we want to prove to ourselves that it's happening :) # 14:29 harry It is probably not a good idea to interpret W3C standards as religious documents. # 14:29 AdamB ... but just not enough. based on experience with microformats # 14:29 Arnaud ACTION: eprodom to extract the examples from the main documents, pick properties and put together a test for those exmaples # 14:30 AdamB ... all these other issues come up # 14:30 harry Its a good idea to make standards that get actual interop and solve problems # 14:30 harry The decentralized social networking problem is a hard problem. # 14:30 tantek abstract conformance criteria ("parsing") is insufficient, we need to take the use-cases that drove these features and get them added to the spec # 14:31 AdamB bblfish: history of rss is 15 years old, there is really important part of formats to support previous versions # 14:31 AdamB ... where backward compatibility is an important part of the work # 14:32 AdamB ISSUE: look at backward compatibility for AS 2.0 # 14:32 jasnell_ backwards compatibility with AS1 is already covered by the core spec doc # 14:33 Arnaud ACTION: eprodrom to extract the examples from the main documents, pick properties and put together a test for those examples # 14:33 trackbot Created ACTION-50 - Extract the examples from the main documents, pick properties and put together a test for those examples [on Evan Prodromou - due 2015-03-24]. # 14:34 AdamB fjh: when will we talk about json-ld # 14:34 AdamB Arnaud: there are bunch of issues open, bunch of new issues that we need to look at what to do with # 14:35 AdamB ... we have limited amount of time so we should discuss prioritizing what we should be talking about # 14:35 harry But given interop in previous Social Web efforts has generally failed so its a good start to have a shared goal/understanding there # 14:36 AdamB ... discussing the agenda and what to do next # 14:38 wseltzer [20 min break] # 14:39 elf-pavlik tantek, please try not to change nesting in items i proposed! # 14:51 elf-pavlik tantek, looks good! since we discussed testing in depth i removed copy of it from proposed list # 15:01 elf-pavlik rhiaro, you could aks AnnB about changing IG telecon times ... # 15:04 bblfish somehow I think I seem to hear better now on talky # 15:04 AnnB I see that note, elf-pavlik and rhiaro # 15:04 AnnB sounds like the IG time is not good for you? # 15:05 AnnB not sure that got recorded correctly # 15:05 wseltzer scribenick: AnnB # 15:05 AnnB hopes everyone will help fill in the scribing .. # 15:05 aaronpk did not know there were two different things that needed to know the scribe nick # 15:06 AnnB Agenda Bashing commences ... fjh and AdamB joined the channel
# 15:06 AnnB elf's link is the static one; tantek's link is the "live" one # 15:06 Zakim On IRC I see AdamB, fjh, danbri, stevenroose, rhiaro, jasnell_, AnnB, SimonTennant, harry, jaywink, bblfish, tantek, tilgovi, Zakim, RRSAgent, Arnaud, cwebber2, pfefferle, # 15:06 Zakim ... elf-pavlik, the_frey, wilkie_, shepazu, KevinMarks, Tsyesika, sandro, Loqi, nickstenn, aaronpk, oshepherd, wseltzer, ElijahLynn, ben_thatmustbeme, mattl, JakeHart, bret, dwhly, # 15:07 trackbot issue-15 -- AS2.0 Vocabulary in many ways duplicates microformats.org and schema.org efforts -- open # 15:07 fjh can we add issue-21 to 11 am session? # 15:07 jasnell_ Considering the test suite: One possible approach forward... Define test cases as Stories. Each Story is about a series of Events that have occurred. An implementation ought to have a limited number of ways of encoding that information as output. The output ought to be able to be fed into a test tool. The tester then asks a specific set of questions against each input for the test. For example: Story: "James posted a blog post at 1:23pm pacifi # 15:07 jasnell_ c. Arnaud liked the blog post and shared it with Tantek". The types of questions the test tool can ask include, "Who posted a blog post?" "Who liked it?" "Who was it shared with?". If the tool is able to extract the correct answer given the input, then the input can be considered to pass the test. To test consumption, an implementation would be expected to transcribe specific inputs into either HTML or Turtle in such a way that the test tool w # 15:07 jasnell_ ould be able to ask the same set of questions and extract the appropriate answers. # 15:07 AnnB W3C team at MIT has lunch together every Tuesday ... and invited us to join them ... we need to get there @noon # 15:09 AnnB fjh = Frederick Hirsch # 15:09 AnnB fjh: wants to understand json-ld topics, relationship to Annotations WG work # 15:10 AnnB Arnaud to Tantek: there are ++ on agenda list; we only have time for ~1-2 topics pfefferle joined the channel
# 15:13 trackbot action-26 -- Tantek Çelik to Review microformats examples in AS2.0 specs -- due 2015-03-17 -- OPEN # 15:13 wseltzer Topic: Microformats # 15:14 jasnell_ took a best guess at the microformats hoping tantek would fix them :-) jeff joined the channel
# 15:15 AnnB tantek: I've captured the changes that need to be made in spec, re: Github issue 83 dret joined the channel
# 15:15 AnnB trying to focus on key items to fix # 15:15 wseltzer AnnB, I'll find the links # 15:15 AnnB patterns of changes are the interesting bit # 15:16 AnnB tantek: in our publishing to date, we're using URLs not URNs # 15:16 AnnB ... I propose we use URLs, because that's the common practice # 15:16 Zakim sees elf-pavlik, bblfish on the speaker queue # 15:16 harry I am kinda waiting for httpRange-14 to make an appearance # 15:17 Zakim sees elf-pavlik, bblfish on the speaker queue # 15:18 tantek Instead of urn:example:person:martin we should use http://martin.example.org/ because that reflects current social web publishing and identity practices, e.g. indie sites and sites like Tumblr # 15:18 AdamB +1 on urls intead of urns # 15:19 harry Does anyone want to support URN usage in the spec? # 15:19 AnnB Arnaud: I suggest we accept this proposal .. but let's have discussion # 15:19 sandro +1 but I note there's semantic confusion possible (differentiating between a person and their site) # 15:19 AnnB ... reminder, we're talking about best practices # 15:19 harry I was waiting for httpRange-14 to rear its ugly and confusing head :) # 15:19 AnnB we can't hear you elf # 15:19 harry Sandro, what's your take? example.org/me#? # 15:20 tantek note that I specifically used username.example.org per modern practices # 15:20 cwebber2 mattl: that's the federation borg's assimilation noise # 15:21 harry I thought it was the ghost of httpRange-14 # 15:21 harry a spectre is haunting the social web, the spectre of httpRange-14 # 15:21 trackbot action-34 -- Pavlik elf to add explaination to the spec about multiple serializations used in examples -- due 2015-02-10 -- PENDINGREVIEW # 15:22 elf-pavlik 3) formal mapping microformats vocabulary to as2.0 vocabulary e.g. owl:equivalentProperty owl:equivalentClass # 15:22 AnnB bblfish: in ontology .. doesn't matter if it's URL or URN # 15:23 AnnB scribing baby ... waaa # 15:23 trackbot action-43 -- Pavlik elf to propose *lightweight* inference based on RDFa Vocabulary Expansion -- due 2015-03-17 -- OPEN # 15:23 harry I'm glad the baby in the background is having the proper reaction. # 15:23 AnnB bblfish: recommends FOAF to identify "person" in various ways # 15:24 AnnB ... already ~10 years of history w FOAF ontology eprodrom joined the channel
# 15:24 AnnB ... I'm against pattern of specifically how the URL should look # 15:24 AnnB evan: clarifies this is only for the examples on AS2.0 document # 15:25 sandro q+ to perhaps summarize the debate very tersely # 15:25 AnnB bblfish: still suggests distinguish between XXX and YYY # 15:25 wseltzer does anyone object to the example changing? # 15:25 Zakim sandro, you wanted to perhaps summarize the debate very tersely # 15:25 AnnB hopes bblfish will fill in the gaps # 15:26 AnnB Sandro: trying to help explain bblfish's point # 15:26 AnnB ... long history in SemWeb community # 15:26 AnnB .... this is called HTTP range 14 # 15:27 AnnB ... I'm personally OK with this example .. but there is a long history with this, and there are people who will complain # 15:27 Zakim sees bigbluehat, harry on the speaker queue # 15:27 AnnB tantek reminds this he's only suggesting this for a specific example # 15:27 Zakim sees bigbluehat, harry on the speaker queue # 15:27 wseltzer doesn't see TimBL in the room # 15:28 tantek Tumblr, Blogger, LiveJournal all support just plain URLs # 15:28 jeff Wendy, were you asking that we invite him? # 15:28 aaronpk just realized "Tumblr" and "Timbl" are very similar # 15:28 sandro timbl *is* planning to join us this afternoon # 15:28 wseltzer was suggesting we not invoke him, though we have invited him # 15:28 AnnB bigbluehat = Benjamin Young # 15:28 bblfish tantek is mistaken, a huge number of foaf files do that AnnB joined the channel
# 15:29 fjh bblfish how can I use foaf currently in a useful manner, most servers/appications seem to be gone # 15:29 tantek bblfish: no company or popular social website does anything with FOAF in their UI # 15:30 sandro harry: I suggest we use best practice for the informative test, examples, ... if people want to use hash-URIs in an appendix... in a spec we should cater to the 99% ... cater to the no-experience web developer # 15:30 bblfish It is pretty widely deployed, and commercial web site of course don't do anything with distributed this # 15:30 harry I think URNs are in general bad practice for anything on the Web # 15:30 AnnB Bigbluehat: switching from URN to URL actually accommodates the HTTP Range 14 issue # 15:30 sandro bigbluehat: using URLs is okay with httpRange-14 # 15:31 sandro +1 agreed, URLs are okay with httpRange-14 BUT Henry's point was about the specific URLs to use # 15:31 harry So my proposal is that if people want to use RDF with hash-URLs, they can do so in an appendix or WG note on other syntax # 15:31 sandro bblfish, I didn't say that. I think you mean "sandro, " # 15:32 harry In fact, an appendix with hash-URLs in Turtle *could* be useful. Augier joined the channel
# 15:32 jasnell_ it's a copy/replace, not that much work, I just don't see it as a big deal # 15:32 AnnB Arnaud -- we're trying to figure out what we can live with # 15:32 tantek -1 to any /#me or anything not popular on the existing social web # 15:32 AnnB bblfish: question is, what do these URLs refer to? # 15:33 bigbluehat fjh next to you :) just not in this WG...so shouldn't be voting... imo ;) # 15:33 fjh bigbluehat oh, vote, vote early, vote often :) # 15:34 harry Yes, feel empowered to file an formal objection # 15:34 AnnB bblfish tries to clarify more about Semantic Web / RDF stuff # 15:34 tantek Arnaud: my proposal was to use username.example.org # 15:34 AnnB Arnaud reins the discussion back in # 15:34 sandro Arnaud: I'm sorry Henry, we're going to go forward over your -1. You can file a formal objection if you want. # 15:34 fjh +1 to moving on even though not in WG # 15:35 AnnB tantek responds to bblfish that this is documented in Issue 83 # 15:35 fjh bigbluehat do you have a reference re using hash to solve Range-14 problem # 15:35 jasnell_ the color of the example url bikeshed is perfectly fine as decided. let's move on # 15:36 AnnB Sandro: to clarify, none of this goes against RDF .. this will all work.... I can explain that later with bblfish, if that'd help # 15:36 jasnell_ there is no normative reference to microformats in the spec # 15:36 jasnell_ the microformats stuff is in non-normative examples only # 15:37 harry For record, I think we should mute off topic speakes # 15:37 tilgovi Is there any problem with using a URL structure that pleases harry et al more? # 15:37 harry There is an open issue and it's been discussed # 15:37 AnnB bblfish: points out that part of confusion is stuff that aren't standards # 15:37 tilgovi It seems to me like this room doesn't particularly care, just that we'd prefer not a URN # 15:38 AnnB harry: the issue of non-standards is being looked at by W3C management # 15:38 harry For future reference, we should mute off topic speakers # 15:38 AnnB tantek: we need to restrain the discussion to the topics on the agenda # 15:39 harry bblfish, when you are willing to talk on topic, please unmute yourself. # 15:39 harry I actually think people who are chronically off-topic should not time to dominate the conversation. # 15:39 AnnB tantek: there are examples inside links and metatags # 15:40 elf-pavlik q+ re: microformat HTML serialization (less important) vs. microformats *vocabulary* (more relevant) # 15:40 wseltzer bblfish, I'm sorry that you're not able to be in the room, but we're trying to proceed on the agenda the group chose # 15:40 AnnB ... instead of using link tag, use "a HREF" # 15:40 fjh jasnell, thought so, just trying to get to know people, thanks # 15:40 AnnB .. instead of metatag use best option # 15:41 Zakim sees elf-pavlik, harry on the speaker queue # 15:41 Zakim elf-pavlik, you wanted to discuss microformat HTML serialization (less important) vs. microformats *vocabulary* (more relevant) # 15:41 bblfish wseltzer: there were a number of discussions regarding all kinds of things that were cut of because a proposal was made that we cannot discuss non published standards # 15:41 AnnB ... if you see a metatag it's probably a mistake, and I've offered alternatives # 15:41 wseltzer bblfish: that's not how the chairs are running the meeting # 15:41 wseltzer they're not taking "blocking" procedures # 15:42 aaronpk tantek: if you see a <link> or <meta> tag inside a div, it's probably a mistake and does not reflect current publishing practices, and i've offered alternatives instead # 15:42 Zakim sorry, harry, I do not know which phone connection belongs to elf-pavlik # 15:42 wseltzer elf, we can't hear you # 15:42 jasnell_ the microformats *examples* are non-normative. the changes to those are purely editorial. I doubt we need to spend much time on them # 15:42 trackbot action-34 -- Pavlik elf to add explaination to the spec about multiple serializations used in examples -- due 2015-02-10 -- PENDINGREVIEW # 15:43 fjh jasnell what do you mean by PR (I keep thinking Proposed Rec) I think you mean problem report? # 15:43 AnnB harry: everyone has their favorites syntax .. for the spec, I suggest the foundation should be JSON-LD, with the other syntaxes in "communities" # 15:44 AnnB Sandro: you can already see that in the tabs # 15:44 bblfish is there even a microformats to rdf mapping? ( there may be but I don't know ) # 15:44 harry Anyways, I'm not 100% happy with the tab approach, but if people want to keep it, go for it. # 15:44 AnnB tantek: I'm concerned about that, to make the meanings consistent # 15:44 AnnB sandro: will we have machine translatability between these formats? # 15:45 AnnB tantek: I think that's why jasnell did all that hard work # 15:45 cwebber2 if the doc needs to be clarified to make clear that the other ones are there for clarity, that's fine # 15:45 AnnB jasnell = James Snell, IBM # 15:45 cwebber2 but making it easy to come in and understad mapping, super great # 15:45 wseltzer 's battery is dying # 15:45 AnnB tantek: microformats are the concrete example I can use to move this forward # 15:45 AnnB ... I'm concerned about # 15:45 bblfish tantek is making a good case for clear semantics +1 # 15:45 AnnB ... about "drift" if we did what Harry proposed # 15:45 Zakim sees bigbluehat, jasnell_ on the speaker queue # 15:46 fjh annotation wg is currently using tabs, for same reasons Tantek mentions, to broaden community and allow clarity # 15:46 AnnB Harry: given that these communities are all active, I can agree # 15:46 harry If the communities become inactive, I would recommend removing them. # 15:46 eprodrom Zakim, mute baby # 15:46 Zakim sorry, eprodrom, I do not know which phone connection belongs to baby # 15:46 Zakim sorry, aaronpk, I do not recognize a party named 'baby' # 15:47 fjh shepazu in document wwhere we show JSON-LD Turtle etc # 15:47 AnnB bigbluehat: cites drupal 7 and RDF # 15:47 sandro ben_thatmustbeme pause buttons are much more important # 15:47 AnnB s/and RDF/and RDFa/ # 15:48 harry My opinion in general is 'alternative syntaxes' are a standardization failure # 15:48 tantek jasnell_: side by side examples will also help with testing # 15:48 harry but that's where we are in this space and all communities are all active # 15:48 AnnB jasnell: in showing multiple formats I'm trying to help people compare side-by-side # 15:48 tantek accomodoating alternative syntaxes seems to be a path forward so let's stick with it # 15:48 sandro harry, it's a "suboptimal solution", not a "failure" :) # 15:49 AnnB thanks bigbluehat # 15:49 AnnB jasnell: we don't need to spend a bunch of time on non-normative examples # 15:49 fjh I hope we can do the JSON-LD right after lunch if we don't get to it before? wseltzer-mobile joined the channel
# 15:49 harry in fact, they cut and paste non-normative examples # 15:49 elf-pavlik we still didn't asnwer what do we do about microformats *vocabulary* (not the HTML serialization) # 15:49 fjh people sometimes look at the examples first # 15:50 elf-pavlik we got stuck on serialization instead discussing vocab (which one can also use in JSON-LD) # 15:51 tantek PROPOSAL: inside <div>s in the examples, use <a href> instead of <link>, and do not use <meta> but use proper visible markup instead such as <img>, <time>, or <span> # 15:51 cwebber2 ... the intent is that it will improve the other formats # 15:51 cwebber2 sandro: not sure I understand enough to know this will be a good idea # 15:51 elf-pavlik we get stuck on serialization instead discussing vocab (which one can also use in JSON-LD) # 15:52 sandro 0 I don't have expertise on Microdata and RDFa to vote on this # 15:52 aaronpk +1 since <meta> inside anything other than <head> seems weird in HTML in general # 15:52 harry I think we discussed testing this morning # 15:52 cwebber2 bblfish: if there's well defined mappings between microformats and rdf, it it's well defined can't it automatically be tested? AnnB_ joined the channel
# 15:53 harry the normative syntax right now is the JSON-LD # 15:53 cwebber2 tantek: there have been automated things, but it's lossy if it's in RDF, because you don't give all the information to the user # 15:53 elf-pavlik we get stuck on serialization instead discussing vocab (which one can also use in JSON-LD) # 15:53 harry but it would be nice to get informative examples all the same # 15:53 cwebber2 ... it's complicated because you want to give all examples of mapping to the user for microformats/etc # 15:53 AnnB_ hmm .. how do I kick my oldself off? or will it just happen? # 15:53 cwebber2 bblfish: this will lead to a problem of extensibility # 15:53 elf-pavlik we get stuck on *not normative* serialization instead discussing vocab (which one can also use in JSON-LD) # 15:53 jeff AnnB it should time out after a while # 15:53 cwebber2 tantek: extensibility is a separate problem, I will declare it out of scope # 15:54 jasnell_ +1 to whatever microformats changes tantek feels may be needed to make the examples better reflect microformats best practices # 15:54 cwebber2 sandro: I think that an rdfa expert needs to weigh in # 15:54 harry If no RDFa experts exist, then we can drop that syntax, but hopefully someone will step up # 15:54 cwebber2 tantek: I didn't touch the rdfa example because of that, I want to avoid improving other examples, but make sure we coordinate # 15:55 cwebber2 Arnaud: I think we need more homework before we can settle this # 15:55 AnnB_ cwebber .. are you OK scribing until lunch? I need to change my plane ticket # 15:55 cwebber2 tantek: okay I won't touch other examples, if someone wants to look at it # 15:55 cwebber2 tantek: I want to get fixes in incrementally so it can be reviewed, if you can let me know if there are outstanding issues # 15:56 cwebber2 jasnell_: keep in mind that only normative format is the json-ld, so any changes that breaks the json-ld serialization, the json-ld one wins # 15:56 cwebber2 jasnell_: yes I'll update, as for the microdata/turtle stuff submit pull requests (???) # 15:56 cwebber2 Arnaud: we're running out of time, and we've barely started scratching this list # 15:57 tantek I'll be submitting incremental pull requests to fix the microformats examples a few at a time # 15:57 cwebber2 harry: I have a proposal we can talk about over lunch, almost all these proposals are some variation are "I'd like to use RDF/etc", but most developers coming here have no idea what rdf is # 15:57 tantek jasnell_: I hope you're able to incrementally accept them # 15:57 cwebber2 ... so for anything that requires inference, type heirarchies, etc, we need to write those down # 15:58 cwebber2 ... a like is a kind of action, put heirarchy (??) into the spec # 15:58 jasnell_ if you can bundle them up as much as possible, that would be helpful # 15:58 trackbot action-43 -- Pavlik elf to propose *lightweight* inference based on RDFa Vocabulary Expansion -- due 2015-03-17 -- OPEN # 15:58 cwebber2 ... make it explicit in the spec, rdf people can see that, others can just follow the spec # 15:59 jasnell_ I will be generally unavailable tomorrow due to conflicting meetings # 15:59 harry PROPOSAL: The spec should simply make any and all spec-processing fully explicit and not rely on any references to RDF, although people can use RDF-based technologies if they so want to implement the spec. # 16:00 dret +1 to harry's proposal. clean layering. have a JSON spec with processing model. have another one for those interested in an RDF view. both camps are happy. # 16:02 dret we only have to make sure if we want to do it that way. the only think we technically HAVE to do per our charter is stay aligned with JSON. # 16:04 elf-pavlik dret, but we can't make LinkedData folks happy if we use some other model, while we can keep RDF hidden from people who just want to use JSON # 16:05 rhiaro elf-pavlik I think we're switching to talky beta after lunch # 16:13 dret bblfish, i completely agree that the charter is not well written. but that doesn't imply that we have to switch to RDF as foundation as the only possible way of fixing that. we can. but we don't have to. # 16:15 dret elfpavlik, i admire how you focus all your energy on making LinkedData folks happy. please just keep in mind that they are a minority, by far. it may not feel like that inside the w3c bubble, but that's a different story. peacekeeper joined the channel
# 16:22 bblfish rdf is just a making explicit of what is implicit in the other formats # 16:22 bblfish so it makes things much faster if you work with tools that are well defined # 16:24 bblfish Otherwise you're back ot the 2004 Atom Syntax discussions - years of discussion on whether something should be an attribute or an xml element, no way of testing things clearly, and so little achieved, ... # 16:39 wilkie rdf is nice. not used by many people, but generally used by the info sciences and meta-data librarians. # 16:40 wilkie maybe not a thing you want to just *use* but something you want to make sure you can support in some way to allow them to use the tools they have developed for defining relationships and graphs # 16:42 elf-pavlik wilkie, RDF has pretty well defined model and many people use it out in a wild (maybe not so much for social networking, yet) # 16:43 mattl elf-pavlik: cause I'm a member. Harry is resolving the issue with systems at w3c... Username conflict or something # 16:43 elf-pavlik IMO unless we see strong reasons why not to reuse it, we should not define new paralel model # 16:44 trackbot issue-17 -- Identity, Agent, Person, Persona, Account etc. need clarifications -- open elf-pavlik joined the channel
elf-pavlik and tantek joined the channel
# 17:13 aaronpk we are waiting on talky until Sandro comes back with his laptop # 17:14 Zakim On IRC I see tantek, elf-pavlik, dret, pfefferle, danbri, rhiaro, jasnell_, SimonTennant, jaywink, bblfish, Zakim, RRSAgent, cwebber2, the_frey, wilkie, shepazu, KevinMarks, # 17:14 Zakim ... Tsyesika, sandro, Loqi, nickstenn, aaronpk, oshepherd, wseltzer, ElijahLynn, ben_thatmustbeme, mattl, JakeHart, bret, dwhly, bigbluehat, rektide, trackbot # 17:14 aaronpk it does client-server connections for >3 users so it's not all p2p and taking lots of bandwidth # 17:14 tantek jasnell_: are you still there? do you want to rejoin the telcon? # 17:14 rhiaro tantek: First item of discussion is whether to keep one or more AS items, because jasnell_ can't be here tomorrow AdamB joined the channel
# 17:15 rhiaro ... The first item specifically requested by Frederick is to summarise the json-ld issue # 17:15 rhiaro ... Specific requests for any other AS items? jeff joined the channel
# 17:16 rhiaro ... We're taking very specific targeted AS items for discussion, before social API # 17:16 rhiaro ... So far we're keeping issue-21, json-ld issue # 17:16 rhiaro ... Are there any other AS issues anyone would like to spend time on? # 17:16 trackbot issue-13 -- Which activity types are built into AS2, and how are they defined/structured? -- open Arnaud joined the channel
# 17:17 trackbot issue-4 -- Do we rely on explicit typing or support implicit typing based on explicit property names? -- open # 17:17 trackbot action-25 -- Evan Prodromou to Discuss social api functional requirement process with ig -- due 2015-01-13 -- CLOSED # 17:17 trackbot action-35 -- Tantek Çelik to Come up with a simple proposal for implicit typing based on property names -- due 2015-02-10 -- OPEN # 17:17 rhiaro ... You can always bring them up tomorrow, this is for while we have jasnell on the phone # 17:18 AdamB elf-pavlik sandro and aaronpk are working on talky # 17:18 rhiaro ... To try to get through more items, if any of the chairs think a discussion is off topic, and nobody disagrees, we're going to move on # 17:18 rhiaro ... If someone wants to raise a separate issue, they can do, but if chairs determine something ot be off topic or a tangent, we're going to declare it and move on # 17:19 rhiaro frederick: What I believe is true about JSON-LD with relation to AS2.0. Not sure I understand all the nuances # 17:19 rhiaro ... My understanding is that JSON-LD is mandatory for AS2.0, and if there's a context in that document it's mandatory, and you cannot replace the context but you could add to it if you need to # 17:20 rhiaro ... and if you need to define vocab that's not in AS namespace, you can use your own space but you need to use an array where you use both your vocab and AS vocab when you create your AS item # 17:20 rhiaro ... So as far as I understand, JSON-LD is good to use, required, no problems. The only issue is how you can modify the context. # 17:20 rhiaro ... This goes without me really understanding the full issue # 17:21 rhiaro elf-pavlik: can you summarise issue-21? (can you hear us?) # 17:21 rhiaro bigbluehat: I read eric's article, seems to be an underlying data model question # 17:21 Zakim sees fjh, eprodrom, jasnell_ on the speaker queue # 17:22 rhiaro ... We ran into a little bit of this with annotation wg and string literal bodies for the body of an annotation, vs resources with identifiers # 17:22 rhiaro ... Otherwise there's no way to say more about a body as a string literal # 17:22 rhiaro ... But json devs don't want objects, they want strings # 17:22 rhiaro ... So currently dealing with that, seems related # 17:22 Zakim sees fjh, eprodrom, jasnell_ on the speaker queue # 17:22 rhiaro ... So would like to hear from elf-pavlik or anyone else about that # 17:22 Zakim sees eprodrom, jasnell_ on the speaker queue eprodrom and fjh_ joined the channel
# 17:23 bigbluehat bblfish tnx for the repost--that's the article I referenced from issue-21. much thanks # 17:23 tantek q+ re: developers just want strings vs. dereferencing objects (complexity problem) # 17:23 Zakim sees tantek, bblfish on the speaker queue # 17:23 rhiaro jasnell_: as I understand it, we have the vocab itself which is defined as an rdf vocab. We also have the syntax, which is JSON-LD harry joined the channel
# 17:24 rhiaro ... right now with the jsonld syntax, the most that is required for an implementor to read an AS doc properly is the json-ld expansion algorithm # 17:24 rhiaro ... there's no link back to that says it has to be rdf # 17:25 elf-pavlik i have issue with SIP connection to zakim, while digitalbazaar SIP works just fine # 17:25 rhiaro ... you can expand it out with the json-ld algorithm, but still handle it with json data without having to do any inferencing or additional processing # 17:25 rhiaro ... there are several activity types listed, eg a Like is a subclass of a Response # 17:25 rhiaro ... But if I parse an AS JSON-LD document, a like and a respond are two separate things # 17:25 rhiaro ...That's just json-ld expansion, you can't necessarily tell that they are related to each other # 17:26 Zakim sees tantek, bblfish, harry on the speaker queue # 17:26 rhiaro ... So key question is: is inferencing based on the rdf model required for implementors? # 17:26 Zakim sees tantek, bblfish, harry, fjh_ on the speaker queue # 17:26 Zakim sees tantek, bblfish, harry, fjh_, eprodrom on the speaker queue akuckartz joined the channel
# 17:26 trackbot action-43 -- Pavlik elf to propose *lightweight* inference based on RDFa Vocabulary Expansion -- due 2015-03-17 -- OPEN # 17:26 rhiaro ... So by parsing an AS doc and getting all the responses, do I have to also get the likes? # 17:26 eprodrom Joining the club # 17:26 rhiaro ... or just things that are explicitly response? # 17:26 harry As said before lunch, we should just be explicit in text about inferencing/sub-class/sub-type issues in spec as we can't guarantee everyone will have a RDF(S) inferencing working. # 17:26 rhiaro ... So that's the key question behind the issue # 17:26 harry It makes the spec longer but worked well for HTML5. # 17:26 rhiaro ... Does someone parsing an AS doc required to do the type inferencing that's included in the rdf vocabulary? # 17:27 Zakim sees tantek, bblfish, harry, fjh_, eprodrom on the speaker queue # 17:27 Zakim sees tantek, bblfish, harry, fjh_, eprodrom on the speaker queue # 17:27 Zakim tantek, you wanted to discuss developers just want strings vs. dereferencing objects (complexity problem) # 17:27 rhiaro tantek: specifically, for the high level question, how much is required of json-ld # 17:27 Zakim sees bblfish, harry, fjh_, eprodrom on the speaker queue # 17:27 rhiaro ... I don't think we should require json-ld, as an implementor tilgovi joined the channel
# 17:27 rhiaro ... The summary point: "developers just want strings" issue is a really important one # 17:28 rhiaro ... For devs to publish and consume this stuff # 17:28 eprodrom Can we say MUST support JSON-LD, MAY support other Linked Data serialization? # 17:28 rhiaro ... I think requiring json-ld and any kind of inferencing adds complexity, will harm adoption, hinder interop # 17:28 Zakim sees bblfish, harry, fjh_, eprodrom, jasnell_ on the speaker queue # 17:28 harry yes, I think that's kinda what we said already eprodrom # 17:28 rhiaro ... As compared to existing state of the art in terms of both closed social web apis, and in terms of minimal indieweb efforts # 17:28 rhiaro ... which have success with achieving interop purely with parsed json results from microformats # 17:28 eprodrom harry: I think that's a happy medium # 17:28 rhiaro ... I can see that having json-ld as a nice to have is fine # 17:29 rhiaro ... but having it required is bad for the spec and bad for chances of success # 17:29 Zakim sees bblfish, harry, fjh_, eprodrom, jasnell_, elf-pavlik on the speaker queue # 17:29 harry exactly, we can't force people to consume *everything*, but we fail as a standards body if we don't say "You MUST support this ONE syntax" # 17:29 Zakim sees bblfish, harry, fjh_, eprodrom, jasnell_, elf-pavlik on the speaker queue # 17:29 Zakim sees harry, fjh_, eprodrom, jasnell_, elf-pavlik on the speaker queue # 17:29 rhiaro ... first, developers in closed systems don't care about interop, json will do perfectly well # 17:29 rhiaro ... when you don't have to follow links from one server to another # 17:30 rhiaro ... but that's necessary if you want to make a distributed social web # 17:30 tantek I will note that indieweb folks are building a distributed social web, peer to peer, that works with "simply JSON" # 17:30 rhiaro ... Also, if you try to build a system where you're mimicking(?) a client to server mode, you lose a lot with just json # 17:30 rhiaro ... This federated API should be somewhat different AnnB joined the channel
# 17:30 tantek thus we've demonstrated you don't need JSONLD for distributed social web across sites & implementations Guest joined the channel
# 17:30 cwebber2 how otherwise will different websites implementing extensions interoperate? # 17:30 rhiaro ... RDF is a unified mechanism for creating a global web of data # 17:31 tantek with microformats and JSON we also produce links you can follow around, WITHOUT JSONLD # 17:31 rhiaro ... If you don't do that, you get what we had in ATOM wg a long time ago # 17:31 rhiaro ... RDF comes with a bunch of tools that you can use wilkie_ joined the channel
# 17:31 cwebber2 how to differentiate between "run" as a verb for joggers, and "run" as a verb for executing a program? # 17:31 rhiaro ... You don't need to require a client to have the inferencing tools # 17:31 rhiaro ... all lwe're doing is specifying consequences of things # 17:31 Zakim sees harry, fjh_, eprodrom, jasnell_, elf-pavlik on the speaker queue # 17:31 Zakim sees harry, fjh_, eprodrom, jasnell_, elf-pavlik, bigbluehat on the speaker queue # 17:31 rhiaro ... eg. there's a relation between a feed and an entry # 17:32 rhiaro ... all of these kinds of things are expressed in english, and you can express in rdf # 17:32 cwebber2 you could spell out the whole URL for those verbs, but # 17:32 cwebber2 "Namespaces are one honking great idea -- let's do more of those! # 17:32 rhiaro ... then you have all the tools you can use to come to various conclusions, like logical flaws # 17:32 rhiaro ... so you get a whole bunch of testing tools # 17:32 rhiaro ... so rdf as a way of specifying what we're trying to express is good # 17:32 rhiaro ... So whether clients are required to do inferencing? # 17:32 aaronpk can someone give me the tl;dr version of what bblfish is saying? # 17:32 rhiaro ... Imagine if you go to someone who doesn't want to inferencing you just have a bad experience, and can go somewhere else # 17:33 bret what happens when incorect data is introduced into that system? # 17:33 harry and also we should probably try to focus on being concise # 17:33 Zakim sees harry, fjh_, eprodrom, jasnell_, elf-pavlik, bigbluehat on the speaker queue # 17:33 rhiaro ... you start moving to getting better experiences as clients start to improve with inferencing. Clients who do will have an advantage # 17:33 bret does that cut the legs off of inferencing? SimonTennant1 joined the channel
# 17:33 Zakim sees fjh_, eprodrom, jasnell_, elf-pavlik, bigbluehat on the speaker queue # 17:33 rhiaro harry: The fact of the matter is that there are people who want to use rdf inferencing. There are many devs who have no idea what that means and will not use it. # 17:34 rhiaro ... We will not get interop if we require something people will not use # 17:34 rhiaro ... *and* we want everyone to use the toolset of their choice # 17:34 bblfish not it does not cut the legs of inferencing: because rdf is built so that inferencing just adds information but never makes anything false # 17:34 rhiaro ... to require interop, we have to spell it out in the spec # 17:34 Zakim sees fjh_, eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud on the speaker queue # 17:34 tantek q+ to point out that our competition is existing deployed open feeds, e.g. RSS/Atom, and simple silo JSON APIs. Anything that adds complexity above that will dissuade developers and win few converts. # 17:34 Zakim sees fjh_, eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud, tantek on the speaker queue # 17:34 rhiaro ... If we want to require any kind of inferencing (basically subtyping and subclassing) we just need to make that explicit in the spec # 17:34 rhiaro ... People can hand code it, or infer it, or whatever # 17:35 Zakim sees fjh_, eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud, tantek, tilgovi on the speaker queue # 17:35 Zakim sees eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud, tantek, tilgovi on the speaker queue # 17:35 rhiaro fjh_: I think we need to be clear about what our hypothesis timbl joined the channel
# 17:35 rhiaro ... I was thinking if you use the jsonld as it is in the document today, it doesn't say anything about inferencing # 17:35 harry PROPOSAL: Any inferencing/sub-class/sub-typing that is required for interop should be spelled out explicitly in the spec. # 17:35 rhiaro ... Currently no requirement for inferencing, but presumably you could do it because it's jsonld # 17:36 rhiaro ... So I don't see what inferencing has to do with this discussion # 17:36 rhiaro ... But if you don't care about it, it has no impact # 17:36 harry then, people can use RDF(S) or handcode JS. # 17:36 rhiaro ... So then the question is - json-ld is json ,so not going to really mess with devs - so why wouldn't you just define the context and require it to be there # 17:36 timbl What you do in the way of inferebcing is not really relevant to the API, yes. # 17:36 cwebber2 I agree that the @context doesn't harm those who don't want it # 17:36 rhiaro ... It doesn't harm anyone, but enables things if you want it # 17:36 rhiaro ... So then we get into what benjamin was saying about string vs object # 17:37 Zakim sees eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud, tantek, tilgovi on the speaker queue # 17:37 Zakim sees eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud, tantek on the speaker queue # 17:37 rhiaro ... So you don't want to add complexity, but I don't see why the context would do any harm # 17:37 Zakim sees eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud, tantek on the speaker queue # 17:37 Zakim tantek, you wanted to point out that our competition is existing deployed open feeds, e.g. RSS/Atom, and simple silo JSON APIs. Anything that adds complexity above that will # 17:37 Zakim ... dissuade developers and win few converts. # 17:37 Zakim sees eprodrom, jasnell_, elf-pavlik, bigbluehat, Arnaud on the speaker queue # 17:37 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud on the speaker queue # 17:37 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, eprodrom on the speaker queue # 17:37 rhiaro tantek: to be clear, our competition is not a green field. There is already existing work. Existing feeds, RSS, atom, AS 1.0 with json, have been implemented across everal systems # 17:38 harry fjh_, so in terms of what features of the RDF data model you are want/use besides inferencing at Annotations, can you tell us? # 17:38 harry Is this just the @context expansion question? # 17:38 rhiaro ... So whatever the hypothetical advantages of rdf/json-ld, the reality is we have deployed working systems and have developers who know how these work, so they're going to need convincing abou tthe extra complexity # 17:38 rhiaro ... The point about a string vs an object - we've had the same problem with microformats # 17:38 harry ah, it's the nested/embedded objects question :) the_frey joined the channel
# 17:39 rhiaro ... We're not going to make progress by arguing from the perspective of accepting architectural baggage # 17:39 rhiaro ... Developers are lazy, already have too much to support # 17:39 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, eprodrom, bblfish on the speaker queue # 17:39 harry My proposal would be let's try to avoid embedded objects unless we can't solve the problem without using them. # 17:39 rhiaro ... If the simplest version isn't a simple as RSS or atom, I think we've failed and developers will ignore it # 17:39 cwebber2 but isn't the simplest version more or less that simple anyway # 17:39 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, eprodrom, bblfish, fjh_ on the speaker queue # 17:39 rhiaro ... If that means we need to come up with a simplified profile, that's fine # 17:39 bblfish I find that developing with good RDF tools makes in fact development much easier # 17:39 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, eprodrom, bblfish, fjh_ on the speaker queue # 17:39 rhiaro fjh_: what is it that makes adding context complicated? # 17:39 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, eprodrom, bblfish, fjh_, timbl on the speaker queue # 17:39 harry That doesn't mean we can't use JSON-LD, it just means that we should try to limit expansion. # 17:40 rhiaro tantek: As a consumer, you want to look at as few things as possible and ignore as much as possible # 17:40 rhiaro ... You just want to display it to the user or whatever you want to achieve # 17:40 rhiaro ... Anyting else we put in, adds more work to dev to dig out info they want to do something with # 17:40 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, eprodrom, bblfish, timbl on the speaker queue # 17:40 rhiaro ... The more we can say here's a simple flat list of objects, the easier it is for devs to do what they want # 17:40 rhiaro sandro: context is just one more property that they're free to ignore # 17:40 rhiaro tantek: yes, we should'nt require them to process it # 17:41 rhiaro sandro: has anyone been proposing that people have to parse the context? # 17:41 harry And *anything* required to process should be explicitly. # 17:41 rhiaro tantek: In the spec json-ld required implies different things to different people # 17:41 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:41 timbl bblfish, when you express how happy you are doing inference, it make s people fea they will be forced to use inferebec engines. # 17:41 rhiaro eprodrom: we have examples in rdfa, turtle, json, mf2 # 17:41 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:42 rhiaro sandro: so we're usin gthe same context every time, so you don't need to understand the context unless you're writing extensions # 17:42 wilkie what I want as an implementor is a solid understandable base that has a clear and comprehendable method of extensibility. AS2 with JSON-LD gets you very far toward this. # 17:42 fjh_ thanks, apologies for clarification question harry # 17:42 harry harry no worries, I thought you meant inferencing # 17:42 rhiaro eprodrom: json is required means we're going to be using json, so yo ushould be looking for this json. If you'd like to store it as turtle, fine (and here's how), but we're going to be talking json # 17:42 elf-pavlik I think you need to just get default language from @context and then you can ingnore it # 17:42 rhiaro sandro: json-ld with a context defined by this group # 17:42 Zakim sees jasnell_, elf-pavlik, bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:42 Zakim sees elf-pavlik, bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:42 cwebber2 yes, a context defined by this group simplifies things enough so you don't *have* to use json-ld stuff unless you want to do extensions # 17:42 Zakim sorry, harry, I do not know which phone connection belongs to jasnell_ # 17:42 tilgovi Isn't it simple enough to say: "Here is our default context. If a client doesn't negotiate for JSON-LD, you _must_ use this context to give them an appropriate JSON response." # 17:43 rhiaro jasnell_: just to clarify what is in the spec currently # 17:43 cwebber2 and that I think is A++, great route, would go that route again # 17:43 rhiaro ... That right now, the spec assumes that the serialization is json-ld document compacted using the normative context document # 17:43 AdamB notices that james like to walk around when he talks :) # 17:43 rhiaro ... What that does is gives us a very (?) json document which does not require json-ld mechanisms to understand it # 17:44 rhiaro ... Everyone should be doing things in a consistent way when it comes to core vocabulary terms # 17:44 sandro +1 jasnell_: compact form JSON-ld using the normative context, this lets one consume the data without understanding json-ld # 17:44 rhiaro ... But when you are parsing, particularly for extensions, it reqiures an extended context to define the additional pieces # 17:44 rhiaro ... So at a minimum, to reliably consume an AS2.0 document, you need to have the json-ld expansion algorithm # 17:44 AdamB cwebber2 - me too :) # 17:44 rhiaro ... That is currently the only json ld requirement # 17:44 sandro jasnell_: but you need to have the expansion algorithm to understand a document with extensions # 17:45 rhiaro ... If the @context is not there because we have a normative @context document, implement can insert that # 17:45 Zakim sees elf-pavlik, bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:45 rhiaro ... If they choose to ignore it and just process the json-ld as regular json, they can do that as well # 17:45 rhiaro ... But that would be optional for the implementor # 17:45 Zakim elf-pavlik, you wanted to discuss using microformats vocab with JSON-LD # 17:45 Zakim sees bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:45 sandro jasnell_: but if developers choose to ignore the context they can # 17:46 Zakim sees bigbluehat, Arnaud, bblfish, timbl on the speaker queue # 17:46 rhiaro tantek: mf off topic for right now, we'll postpone. Is that okay? # 17:46 harry So in terms of JSON-LD expansion, it is only required when extensions are used, correct? # 17:46 Zakim sees Arnaud, bblfish, timbl on the speaker queue # 17:46 harry My guess is that implementation-wise, there will be lots of ignoring of @context amongst non-RDF developers. # 17:46 rhiaro bigbluehat: This is what amounts to our resolution to the string body issue (annotations) # 17:46 rhiaro ... this is the most reasonable solution from previous discussion # 17:46 rhiaro ... We'll have two property options in default @context # 17:47 harry So if any interop is required, we need to explicitly require it and not expect end-users to have JSON-LD conformant processing by default. # 17:47 rhiaro ... one for string literal, one for richer graph # 17:47 Zakim sees Arnaud, bblfish, timbl, harry on the speaker queue # 17:47 rhiaro ... that lets yo umake statements about the body # 17:47 rhiaro ... People wanted to provided csvs as annotations (which got awkward) # 17:47 rhiaro ... So we're saying you can do the string and the object, it's deferred to how that's expressed in the @context # 17:47 tantek interesting, we ended up with a similar solution with microformats JSON - you can get a string, or potentially an object body # 17:47 rhiaro ... We're providing default context for json-ld. You can just use key names if you're using json # 17:47 Zakim sees Arnaud, bblfish, timbl, harry on the speaker queue # 17:47 Zakim sees bblfish, timbl, harry on the speaker queue # 17:48 rhiaro ... The first, to do with inferencing - I hope we put that to bed # 17:48 rhiaro ... But just information, for LDP we do not depend on inferencing # 17:48 bigbluehat tantek: do you have a link to the microformats JSON string/object thing? # 17:48 jasnell_ can we please have a formal proposal on whether or not inferencing is required # 17:48 rhiaro ... We're not going to require clients or servers to do inferencing to understand LDP # 17:48 rhiaro ... SO we just have to agree that's what we're doing # 17:48 Zakim sees bblfish, timbl, harry, eprodrom on the speaker queue # 17:48 rhiaro ... The other is to do with the cos tof making this json-ld rather than json # 17:48 jasnell_ specifically.. a proposal that inferencing is not going to be a requirement # 17:48 rhiaro ... From a consumer point of view, if you want to look at it just as json you ignore the context # 17:48 rhiaro ... If you know what's in there because of out of band information, that's okay # 17:49 rhiaro ... Becaus enot only do you have to provide context, you also have to make sure you're sending consistent data # 17:49 rhiaro Arnaud: which is why you have to acknowledge that # 17:49 Zakim sees bblfish, timbl, harry, eprodrom on the speaker queue # 17:49 rhiaro ... Every json with a @context doesn't just become json-ld # 17:49 rhiaro ... We're actually raising the bar for the producer # 17:50 rhiaro timbl: raising the bar in the sense that the @context has to make sense # 17:50 rhiaro ... So just clarifying Arnaud, more work on the producers side, you're talking about having them put on a @context, will that context be a constant piece of text that's produced by this wg? # 17:50 rhiaro ... Or do they have to be more careful about the json they use that it matches the context? # 17:50 rhiaro Arnaud: The selling arguement for json-ld is 'just add a context' but there's more than that # 17:51 rhiaro ... whether this is significant or not, I leave that to implmeentors # 17:51 rhiaro bigbluehat: eg. you can't have lists of lists # 17:51 rhiaro ... json-ld can't reference items in lists of lists # 17:51 Zakim sees bblfish, timbl, harry, eprodrom on the speaker queue # 17:51 rhiaro ... We ran into some issues with open annotations data model # 17:52 tantek I'm letting folks that don't usually speak, speak-up in person in deference over the queue right now. # 17:52 rhiaro ... The nesting structure of the json has to map to resources # 17:52 tantek including Annotation WG folks who we don't normally get to hear from # 17:52 rhiaro ... you're not mixing random properties from different contexts # 17:52 tantek chairing to explicitly encourage voices that we don't hear from as often # 17:52 rhiaro timbl: If this group decided they wanted to make json in a particular form, which you might think is an ontological stretch, then you can have a special rdf property to match it to # 17:53 rhiaro randall: If the group were to call a certain key part of the standard ontology, producers might have to reshape their data to put a context on it # 17:53 rhiaro eprodrom: the bigger problem is more extension types # 17:53 rhiaro ... We've defined in our vocabulary a Like activity type # 17:53 tilgovi randall == me (tilgovi) # 17:53 AdamB we call them bumps instead of likes # 17:53 rhiaro ... and your application decides to call it 'floop' instead of 'like' and there's some subtle semantic difference peacekeeper joined the channel
# 17:53 Zakim sees bblfish, timbl, harry, eprodrom on the speaker queue # 17:53 Zakim sees bblfish, timbl, harry, eprodrom, fjh_ on the speaker queue # 17:54 rhiaro ... the question is what burden is on conumsers ... if they get a million likes a day and four floops, will they see floops as likes, or discard them # 17:54 cwebber2 zaphod, now that's a floop who knows where his towel is # 17:54 rhiaro ... Seems unfair to floop.com... The idea might be that we put some burden on the consumers, yo uhave to make best efforts to figur eout what that floop is. I fyo ufind out it's a like, you put it in your like bucket # 17:54 rhiaro ... Or on producers - if you are producing these, they might get lost # 17:55 fjh_ you could alias floop to like in JSON-LD, no? # 17:55 rhiaro sandro: do yo uneed a different context for extensions? # 17:55 Zakim sees bblfish, timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 17:56 rhiaro ... if extensions ar epresent, consumers have to understand json-ld # 17:56 cwebber2 eprodrom: isn't it true that someone can still do a "floop" or a "fave" outside of json-ld anyway # 17:56 rhiaro jasnell_: If I have an extension property floop, I just do key name floop and value 1. The json-ld expansion process will just drop that property without a context # 17:56 rhiaro ... In order to keep that from being dropped and not have to put it in a context, I'd have to make that key name be a URL # 17:57 bblfish so one can ask people who write extensions to write them the right way # 17:57 rhiaro ... If I had an extension property, there is either a very specific way to represent it - as a URL - or declare it in an @context that the extension algorithm can use # 17:57 rhiaro tantek: I think this is off topic because nobody is questioning if you can do extensibility with current spec # 17:57 Zakim sees bblfish, timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 17:57 rhiaro ... There may be some details, but not relevant to issue in particular # 17:57 rhiaro ... That's why I want to close off that avenue of discussion # 17:57 eprodrom cwebber2: it'll just be annoying # 17:57 Zakim sees bblfish, timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 17:58 Zakim sees timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 17:58 rhiaro bblfish: I was misunderstood. Most of the rdf tools I use, I don't do any inferencing in client # 17:58 eprodrom cwebber2: if in pump.io, you posted a "floop" to someone, it wouldn't add the actor to the object's "likes" # 17:58 harry fjh_, what precise part of the JSON-LD @context expansion algorithm do you need? # 17:58 rhiaro ... Do a little of hand aided inferencing, we don't have tools # 17:58 harry i.e. do you need the whole algorithm to run? # 17:58 rhiaro ... But pretty much in the camp of how far can I go with no inferencing # 17:58 Zakim sees timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 17:58 rhiaro ... So the second thing there is, when people speak about tools, I think next year the tools are going to be advancing quickly in rdf space # 17:59 rhiaro ... what might be difficult for peopel to do now might be easy in a few months time # 17:59 harry Is always possible tools will expand in the future but let's not bet the spec on it. # 17:59 rhiaro ... so we shouldn't worry about 'difficulty' of things # 17:59 fjh_ harry, personally what I'm concerned about is enabling annotation to work with social web and reasoning when applications can benefit from the combination # 17:59 Zakim sees timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 17:59 cwebber2 eprodrom: maybe the right route is, json-ld or not, assume that if a type is different (even if subtype), you can't expect it works just the same way # 17:59 harry OK, so we're not going to require reasoning. # 17:59 elf-pavlik can we do - with no extensions just use normative @context and treat it as JSON, if you want to extend you need to follow JSON-LD requirements? # 17:59 harry Otherwise than being explicitly listed I htink. # 17:59 harry So fjh_, what part of extensibility do you need? # 17:59 rhiaro ... I agree with the extensbility thing. If people want to write extensible json-ld, they should do it the right way. Perhaps that should be part of the spec: if you want to extend, make sure it's rdf so information not lost # 18:00 harry You just want to mix in annotations with a different namespace, right? # 18:00 fjh_ right, but we don't want to preclude the benefits of JSON-LD while recognizing tank's concerns # 18:00 harry That's fine, people who like those benefits can use them # 18:00 rhiaro ... And then, I don't do any inferencing, having properties such as rdfs: domain, range, etc, is still useful. Not for clients, but for helping you think about what you're doing # 18:00 eprodrom Is the wiki down, or it just me? # 18:00 rhiaro ... When you're using domains and ranges, it helps to come to consensus much faster. You can see the consequences. # 18:01 rhiaro ... Then you build a system that is extensible, built on good principles, open world assumption # 18:01 Zakim sees timbl, harry, eprodrom, fjh_, cwebber on the speaker queue # 18:01 Zakim sees harry, eprodrom, fjh_, cwebber on the speaker queue # 18:01 rhiaro timbl: You have to acknowledge different communities of developers # 18:01 rhiaro ... When you make declarations like 'developers want strings', there are groups who want different things # 18:02 Arnaud bblfish: you really need to make your comments shorter - I know it's not easy but we can't afford to have people speak for so long # 18:02 rhiaro ... I think json-ld looks like a place that allows rdf people to see an rdf view, and json people to see a json view, without getting in the way of each other # 18:02 rhiaro ... The reason people like json is it happens to match the structure of languages they like # 18:02 rhiaro ... And with RDF, if everything is turtle, you can concatenate everything together like you can't in JSON Guest joined the channel
# 18:03 rhiaro ... There are reasonable reasons for liking both # 18:03 rhiaro ... You can do different things with different things # 18:03 rhiaro ... RDF has a longer term extensibility story # 18:03 Zakim sees harry, eprodrom, fjh_, cwebber on the speaker queue # 18:03 rhiaro ... So the answer to this issue... how to close this issue: get a few principles down # 18:03 bblfish but yes, 100% agree with timbl that the meshing of data is the most important and simplest use case for rdf # 18:03 rhiaro ... The idea is we're going to support both communities # 18:04 rhiaro ... It's going to be really simple for anyone to read it into an rdf processor, a json processor, json people can ignore context completely # 18:04 rhiaro ... If there really are possible misunderstandings about what happens if you ignore hte context they should be exlpained # 18:04 rhiaro ... You could have 75% people in the group looking at is a json exercise # 18:04 rhiaro ... If you pick a key like foaf:name, thing sthat are reall ywell known in rdf # 18:04 rhiaro ... you can have people workin gon the mapping # 18:04 Zakim sees harry, eprodrom, fjh_, cwebber on the speaker queue # 18:04 Zakim sees harry, eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:05 rhiaro ... The technique is to use JSON-LD to serve both communities as well as possible # 18:05 Zakim sees harry, eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:05 rhiaro ... With reading data I think it's going to be as easy as writing data... dont' know if it will work as well with both communities # 18:05 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:05 rhiaro ... The RDF community will use that context to expand to RDF timbl joined the channel
# 18:06 rhiaro ... So our spec accepts that reality. How does that impact you? # 18:06 rhiaro ... From annotations perspective, you're going to put another context in there # 18:06 tantek eprodrom could you reload wiki? works for me now # 18:06 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:06 rhiaro ... RDF people will expand that, json people will use it as a string # 18:06 rhiaro ... Is there any part of the json-ld algorithm that we need to require people to process? # 18:07 rhiaro ... We should encourage them, but concerned that we can't require that # 18:07 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:07 rhiaro ... is there anything we should require to to make AS work for json? # 18:07 rhiaro ... What should a non rdf json view do to make things work? # 18:07 rhiaro ... if you care about rdf you get them, if you don't you don't eprodrom_ joined the channel
# 18:07 harry We will require a @context, but we should be aware that additional processing/constraints will simply not be followed by non-RDF aware developers # 18:07 rhiaro ... it's normative in the spec to have a json-ld expansion algorithm # 18:08 harry Thus, we have to explicitly call out anything that is *super* important # 18:08 rhiaro ... We agree that you as a revceiver you can ignore the @context if yo udon't care # 18:08 rhiaro ... only thing I'm not clear on is whether the context should be there or not # 18:08 harry Otherwise, JSON folks who aren't using a JSON-LD parser (i.e. the vast majority) won't do that processing step. # 18:08 rhiaro ... There will be a @context defined normatively in AS2.0, and a JSON-LD serialization that will be normative # 18:08 rhiaro tantek: My understanding: because we've defined the context in the spec, publishers don't actually have to put the context in their AS output # 18:08 harry If it's missing the context then we get the context via the media-type # 18:08 rhiaro ... If it's missing a context, it comes from the spec # 18:09 rhiaro fjh_: If we know there's a reliable context, that's my primary concern # 18:09 harry We do need a media-type/content-type for AS2.0 # 18:09 bblfish do you need a sperate mime type for clients to know when they are getting AS2 json that is not json-ld? # 18:09 harry but we already have that issue and will be registered with IETF once we are done baking spec # 18:09 rhiaro ... If you recieve a document with that content-type, you know which spec it's deifned by, which tells you default @context # 18:09 rhiaro tantek: there are other issues, but that is resolved # 18:10 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:10 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:10 harry I think the other big issue was requiring inference # 18:10 rhiaro eprodrom: With current issue, how do we close it? # 18:10 rhiaro tantek: did we close enough of your [fjh] questions? # 18:10 harry No one is 100% happy but it's important to get the communities having interop. almereyda joined the channel
# 18:10 bblfish I think for extensions clients that need to read it, could be required to use inferencing # 18:10 rhiaro fjh_: are you able to use json-ld and @context if you want to. I think the answer is yes, that should be possible # 18:11 rhiaro harry: if context isn't explicit, it's implicit # 18:11 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:11 Zakim sees eprodrom, fjh_, cwebber, jasnell_ on the speaker queue # 18:11 Zakim sees eprodrom, cwebber, jasnell_ on the speaker queue # 18:11 rhiaro tantek: since we've resolved the aspects of this issue you've brought it up, we leave the remainder open for this group, but not at expense of the rest of today's agenda # 18:11 rhiaro eprodrom: a long term issue that we'll be dealing with # 18:11 rhiaro ... If we address AS2.0 as it is, we can defer this # 18:12 wilkie I'm very excited for json-ld, proper support for extensibility, and rdf! # 18:12 rhiaro fjh_: we can record that this part of the issue is no longer an issue # 18:12 rhiaro tantek: still other aspects to the issue, so can't close it. But can postpone # 18:12 Zakim sees eprodrom, jasnell_ on the speaker queue # 18:12 rhiaro ... Anyone else on the queue has something to add, or we can move on? # 18:12 harry eprodrom, does explicitly spelling out inference in spec help with *known* vocabularies? # 18:13 harry In terms of unknown vocabularies and extensibility, people will probably just have to use RDF # 18:13 rhiaro cwebber2: so context thing is resolved, we should close issue # 18:13 rhiaro PROPOSAL: Please split issue-21 into sub-issues pdurbin joined the channel
# 18:14 sandro PROPOSED: Close ISSUE-21 re-affirming we'll use JSON-LD as in the current draft, with a normative context, and let people bring up sub-issues # 18:15 harry would be lovely to move to Social API discussion # 18:15 rhiaro RESOLVED: Close ISSUE-21 re-affirming we'll use JSON-LD as in the current draft, with a normative context, and let people bring up sub-issues # 18:16 trackbot Issue-4 -- Do we rely on explicit typing or support implicit typing based on explicit property names? -- open # 18:16 rhiaro eprodrom: can we close this? unless tantek wants to complete action-35 # 18:16 trackbot action-35 -- Tantek Çelik to Come up with a simple proposal for implicit typing based on property names -- due 2015-02-10 -- OPEN # 18:17 rhiaro ... we'd have to do some serious work to take out explicit typing # 18:17 rhiaro tantek: since october we've gained more experience with different types of posts # 18:17 rhiaro ... It is working even when we add more types, we don't need explicit typing # 18:17 rhiaro ... Data gathered since october f2f says we don't need explicit types # 18:17 rhiaro eprodrom: if we're not closing it, we don't need to discuss it # 18:18 rhiaro ... Demos of API candidates showing user stories # 18:19 rhiaro ... We just spent several weeks of folks writing user stories, and then another week of voting, then ben_thatmustbeme volunteered to go through and cluser user stories by consensus # 18:19 rhiaro ... There are a bunch with controversy, and some with consensus against # 18:19 rhiaro ... What social web means is different to different people # 18:19 rhiaro ... User stories has beena good way to constrain that # 18:19 rhiaro ... So we have some time for people to demo implementations of existing candidates # 18:20 rhiaro ... various, including micropub, pump.io, ldp and others # 18:20 rhiaro ... For the demos, you demonstrate a user story, not something random # 18:20 rhiaro ... And you do so using an implementation of one of the candidates # 18:20 rhiaro ... So does anyone have any issues with those constraints? Guest joined the channel
# 18:21 rhiaro eprodrom: If we're talking about candidates in terms of what we take to recommendation, demos are great, but would love to see written proposals that say this is what the API standard is, rough overview, this is what will become the document that we edit fjh joined the channel
# 18:21 rhiaro tantek: I'm pretty sure all the candidates have documents like that already # 18:21 rhiaro ... If you're asking for better summaries, separate request # 18:21 rhiaro tantek: part of the reason, there are a lot of proposed candidates, want to use rough consensus and running code filter # 18:22 rhiaro ... let's use that as a way to see what peole have working, and narrow the field # 18:22 rhiaro ... we don't pick one today, but see if there are a couple that we can choose among, or produce some sort of hybrid or something # 18:22 rhiaro ... so for each one of these demos, people must stick to user stories # 18:22 rhiaro eprodrom: are we ruling out proposals that have not been implemented? # 18:22 rhiaro tantek: we're starting to narrow the field among what's been proposed # 18:22 rhiaro ... if something has been proposed but someone isn't demonstrating it, don't want to spend a lot of time talking about it # 18:23 rhiaro ... We can even have multiple working drafts on the same subject # 18:23 rhiaro ... So given those requirements, lets do demos # 18:24 rhiaro ... This is an app running on it's own domain # 18:25 rhiaro ... first sign in, like oauth, asking for scope to post # 18:26 rhiaro ... I signed in. My website granted this app an access token that the app can use to post to my site # 18:27 rhiaro ... makes a POST request to website using form fields, and access token # 18:27 rhiaro ... Proof of concept (that's why it has debugging info) # 18:28 bret elf-pavlik++ thanks for the working link # 18:28 rhiaro ... Site responds to micropub request with location it is posted at # 18:28 rhiaro ... Mobile app would do the same, but may want to render post itself. Could go fetch location of post and do what it wanted # 18:29 rhiaro AnnB: so that's user posts a note user story? # 18:29 rhiaro ... Different domain, totally unrelated t oQuill # 18:30 rhiaro ... These are types that only my website and this app know about # 18:30 rhiaro ... Posted coffee with one click, chose from a list # 18:31 rhiaro ... No confirmation for this one, more user friendly # 18:31 rhiaro wants to use teacup, need to get micropub working # 18:32 rhiaro ... Next demo: posting a file, specifically a photo # 18:33 rhiaro ... For new users, prompts you to attach your instagram # 18:33 rhiaro ... Uses instagram as a client for posting photos # 18:34 rhiaro ... So any photos you post on instagram does a real time micropub post to your site # 18:34 rhiaro ... *aaronpk opens instagram on phone, takes photo* Guest joined the channel
# 18:35 elf-pavlik post post a note and post a file both have edit (replace) and delete part # 18:35 rhiaro ... Just like a plain text note, same properties, but also multipart encoded file # 18:35 rhiaro ... So instagram can disappear and file will still be there. Instagram is just transport # 18:36 rhiaro ... Could also use private instagram as transport # 18:37 rhiaro ... My site supports photos and videos, but not other filetypes # 18:37 rhiaro tantek: does quill or teacup support updating or deleting? # 18:37 rhiaro ... but ben's micropub client and sever support edit and delete # 18:38 rhiaro eprodrom: aaronpk, micropub is about posting new activities # 18:38 rhiaro ... things like traversing your social graph and updating profile are not things it supports? # 18:39 rhiaro aaronpk: micropub intentionally doesn't support any reading operations # 18:39 rhiaro ... so to read the site, you'd parse the microformats in the html # 18:39 elf-pavlik Q: where do we find details about *extending* activity types? # 18:39 rhiaro ... micropub is everything you can't do with get requests # 18:39 rhiaro ... it wouldn't be a stretch to update profile information, we just haven't done that yet fjh joined the channel
# 18:40 rhiaro eprodrom: so if we were to put together a social api based on micropub, we would assemble it together with read stuff # 18:40 rhiaro ... Also about publication technique. What happens if micropub becomes a w3c standard # 18:40 bblfish oops I have to go to another teleconf in 20 minutes # 18:40 rhiaro ... We could stabalise it and make it standard, or we could skip over it # 18:41 rhiaro ... up to the group. aaronpk is editor of micropub spec, so also what aaronpk wants # 18:41 rhiaro ... in terms of what w3c needs, it needs a submission, which is a defined w3c process, and that document becomes something we work on # 18:41 rhiaro eprodrom: the community process around activitystreams is gone or almost entirely gone # 18:41 rhiaro tantek: so it made sense for it to find a new home # 18:42 rhiaro ... if there's a flourising community, sometimes it doesn't make sense to fork the spec # 18:42 rhiaro ... w3c could normatively reference a spec instead, different requirements # 18:42 bret we should use semantic versioning on specs ;) then we can normatively reference a tag of the spec # 18:42 rhiaro ... We could build social api spec by incorporating by reference, and adding more features # 18:43 AnnB asks bblfish if I could jump him in the q for a comment related to this convo? # 18:43 rhiaro ... It's a two way dialogue. If there are user stories we want to support, we can suggest adding it to micropub # 18:43 rhiaro AnnB: An advantage in corporate world, it's easier for us to use a spec that's endorsed by w3c # 18:44 rhiaro ... the more formal it is, the easier it is to use # 18:44 rhiaro tantek: our charter says we're going to produce three things # 18:44 rhiaro ... even if one of them says 'go implement this' # 18:44 rhiaro ... if someone else already has a spec that satisfies charter # 18:44 rhiaro ... that's up for discussion, but that's why we're here # 18:45 rhiaro AnnB: How we anticipate using a lot of this technology is leaning on our vendors/suppliers to say 'use these w3c standards' # 18:45 tantek elf - I think the micropub spec says how to make requests / file issues # 18:45 AnnB in our dreams, our suppliers would listen to us! hehe # 18:45 harry Just going to note that we'd like to hear from people about other social sites/projects/companies/products interested in this API # 18:46 bret elf-pavlik: I would come up with a specific use case that requires nested data and propose that on the wiki and then ping people # 18:47 mattl bblfish: you're looking at a blank screen. # 18:47 mattl bblfish: one sec, we're plugging a laptop into the projector. # 18:48 rhiaro ... Returns html and javascript, which fetches rdf, finds out it's a foaf profile, fetches other profiles # 18:49 rhiaro ... which are also foaf profiles, all over the web # 18:50 tantek this is video playback right? recording from a year ago? # 18:50 rhiaro ... The aim is to show that with javascript following rdf around the web you can create something that looks like facebook. This was done by a couple of engineers who didn't know about rdf # 18:51 rhiaro ... Shows it's not difficult to write with the right tools # 18:51 rhiaro ... Uses LDP, but you don't need LDP to follow resources around web # 18:51 tantek video finished and YouTub started auto-playing something else LOD2? # 18:52 rhiaro eprodrom: didn't see activity streaming, was social graph navigation, which is important, but have you thought about activitystreams style interface? # 18:52 rhiaro bblfish: a foaf:blog relation could point to an activity stream # 18:53 rhiaro ... then find friends' activity streams and aggregate them to a wall # 18:53 rhiaro eprodrom: so something like an inbox where all your friends stuff comes to you # 18:53 rhiaro bblfish: up to client to decide what kind of view to generate # 18:54 rhiaro AnnB: Kevin says is there a URL that is not a video # 18:54 tantek KevinMarks asks if there is a URL that is not a video # 18:54 harry put it in IRC either a live site or a code link # 18:55 rhiaro tantek: URL where results of demo can be viewed, not just code # 18:56 tantek is there a URL where we can see what the demo was showing? # 18:56 AdamB question to bblfish, they are looking for a public URL that goes to that working demo # 18:56 harry its cool if it's not live yet, just ping us later when that goes up # 18:57 rhiaro tantek: next demo, ben_thatmustbe me, Woodwind as client and Postly as server that supports micropub # 18:58 bblfish really sorry, I have a conf call that I need to go to to raise funds so that I can implement much faster the specs that we are discussion here # 18:59 rhiaro ben_thatmustbe: This is an inbox that pulls other people's posts in a feed # 18:59 rhiaro ... It directly posts to my site, again with micropub # 18:59 rhiaro ... Which has a reply to link, saying what it's a reply to # 18:59 rhiaro ... Will show up on Aaron's post in about 5 minutes # 19:00 rhiaro ... once every 5 minutes, webmention sent to posts I've mentioned # 19:00 rhiaro ben_thatmustbe: This also allows liking or starring # 19:01 rhiaro ... after 5 minutes, aaronpk will recieve webmention and can display the like however he likes # 19:01 AdamB q+ what about if you want likes to be anonymous # 19:01 Zakim AdamB, you typed too many words without commas; I suspect you forgot to start with 'to ...' # 19:02 rhiaro ... Woodwind has create. My client on my site when I log in has delete # 19:03 rhiaro ... will send another webmention to aaronpk's site to say it's deleted # 19:04 rhiaro ... Because people are afraid to support maybe politically charged things # 19:05 rhiaro AnnB: Likes are anonymous but all Posts have your name on # 19:05 AdamB thanks tantek for the link # 19:05 rhiaro ... Except that there was a situation where people did not want to post controversial things # 19:06 rhiaro ... So had to set up a separate site for anonymous posts # 19:06 rhiaro tantek: do any of the current user stories cover those scenarios? # 19:07 rhiaro tantek: in general, please fork user stories and add to More page # 19:07 rhiaro ... Next demo: ben_thatmustbe does generic social network client # 19:07 rhiaro ben_thatmustbe: Wrote own little micropub app # 19:08 jasnell_ I'll turn around and pretend I can still see it. Would that help? # 19:08 tantek using Android device, running an Apache Cordova app compiled to Android # 19:08 rhiaro ... written in html/css/js then compile for android # 19:08 tantek and showing it on projector via Photo Booth app # 19:09 rhiaro ... can choose where server sends it to, including facebook and twitter # 19:09 rhiaro ... I ask my site what I support to syndicate to # 19:10 rhiaro aaronpk: it's specifically asking 'give me list of synidcation targets' # 19:11 rhiaro ... you'd have to be able to selectively hide parts of profile if that's part of profile # 19:14 rhiaro ... at that point it's now processing in the background Arnaud joined the channel
# 19:15 rhiaro ... it went from tantek's url, parsed microformats -> json structure # 19:16 jasnell_ I have to drop now. might be able to jump back on later # 19:17 rhiaro ... actually no micropub in that, just microformats harry joined the channel
# 19:18 rhiaro ... What it doesn't do is save the fact that I liked it in the interface # 19:19 rhiaro ... Should show me that I've already liked this post # 19:19 rhiaro ... Easy for things liked using Monocle, but if I've liked something with another app that's harder # 19:19 harry sounds like a fascinating version of the Salmon Protocol issue, likely a bit easier to solve # 19:20 rhiaro ... I want to send tantek a notification, as a homepage webmention # 19:20 rhiaro ... which you can use to make a list of your followers # 19:23 rhiaro cwebber2: combine showing pump.io and mediagoblin # 19:24 rhiaro ... You can see recent activities on the side # 19:24 tantek shows logged in identica home page with personal reading timeline Guest joined the channel
# 19:27 rhiaro ... Go to Meanwhile tab, this happens across two activities # 19:27 rhiaro ... First note just submitted to followers, second public # 19:28 rhiaro ... Also in pumpa, interface to social graph. Shows Activities # 19:29 rhiaro ... Jessica works on a libaray called (?) that makes it easy to build applications # 19:30 rhiaro Jessica: at the moment it's mainly developed for clients, but hoping to expand that # 19:32 rhiaro ... Now to local development instance of mediagoblin # 19:32 rhiaro ... Mostly for submitting images, audio, video etc # 19:33 rhiaro ... can use images submitted with mediagoblin that work for pump.io # 19:34 rhiaro ... Right now we have support for API side of things # 19:36 rhiaro tantek: the pypump demo posted to server? permalink? # 19:37 rhiaro Tsyesika: posted a note on a public list, so should be able to see it # 19:38 rhiaro eprodrom: are there parts of the pump.io api not implemented? # 19:39 rhiaro Tsyesika: haven't implemented likes or (?) yet # 19:39 rhiaro Tsyesika: if you send a share nothing happens # 19:39 rhiaro eprodrom: you mentioned web socket interface? it's sloppy in pump # 19:39 rhiaro ... Last question, have you tried mediagoblin with any android clients? # 19:40 rhiaro cwebber2: we'll have a summer of code project for making a mediagoblin uplload client # 19:49 Zakim Attendees were +1.617.715.aaaa, confroom, elf-pavlik, Sandro, bblfish, jasnell, bret, KevinMarks timbl, tantek and Guest joined the channel
Arnaud and fjh joined the channel
# 20:00 elf-pavlik bblfish, how to do ldp:hasMemberRelation but for inverse direction? eprodrom and harry joined the channel
# 20:04 rhiaro tantek: agenda page updated to list every demo that happened # 20:05 rhiaro ... special thnaks to cwebber2 and Tsyesika for jumping in with unplanned demos # 20:06 Loqi I added a countdown for 3/17 6:30pm (#5654) # 20:06 aaronpk !meme one does not simply make dinner plans in Boston # 20:07 rhiaro aaronpk: that was so fast! You had that ready? Arnaud joined the channel
# 20:11 rhiaro ... Any other social API issues peole want to add? # 20:11 elf-pavlik sandro it sounded better before you connected this other camera or something timbl and Guest joined the channel
# 20:12 elf-pavlik can you disconnect this device you connected to laptop with talky.io ? # 20:12 rhiaro ... We have five items, lets see if we can get through them SimonTennant joined the channel
# 20:13 rhiaro eprodrom: Is someone actually goign to work on a proposal for all of these, eg. hydra? # 20:13 elf-pavlik can someone disconnect this black device that sandro connected to laptop with teleconf? # 20:14 rhiaro ... From the demos, it is clear that there are two actively developed, well supported social APIs that have drawn strong amounts of interest # 20:14 rhiaro ... My proposoal is the working group narrow the candidates for APIs under our consideration down to those two, micropub and pump.io # 20:14 rhiaro ... because we saw user stories we agree on using multiple clients # 20:15 rhiaro harry: the proposal that has been left out that wasn't demoed was LDP, which is broadly compatible # 20:15 rhiaro ... sandro understands LDP, it would be premature to throw it out right now # 20:15 rhiaro tantek: I think we should. The user story henry demonstrated was not even near the top of stories that had any consensus # 20:16 rhiaro ... The other two candidates had numerous user stories that had support # 20:16 elf-pavlik ben_thatmustbeme can you disconnect this black device from laptop that runs talky? # 20:16 rhiaro sandro: I did a demo a year ago of posting with LDP. Some stuff not finished, so didn't want to demo. People have done stuff with LDP, but aren't here # 20:17 rhiaro AnnB: feels prejudiced. I'm not sure I should have an opinion because I@m not a dev, but it's important to listen to this point # 20:17 rhiaro tantek: how to evaluate how far advanced candidates are? # 20:17 elf-pavlik ben_thatmustbeme does it have some device connected with cable? USB i guess ... it just makes lots of noise and worked beter without it # 20:17 rhiaro AnnB: There's a larger representation of one community here than another, so of course that's where the demos came from # 20:18 elf-pavlik not the frame, some device that sandro connected in the end, makes noise :( AnnB joined the channel
# 20:19 rhiaro ... but there might be convergence, wouldn't take LDP off the table just yet # 20:19 elf-pavlik i've seen sandro connecting some black device with usb cable to that laptop and then it got noisy # 20:19 AnnB gee .... is rhiaro still having to scribe? # 20:19 rhiaro sandro: demos were impressive, but there are a whole lot of technical issues we haven't talked about at all # 20:19 Zakim sees fjh, eprodrom, harry on the speaker queue # 20:19 AnnB seems like someone else should step up to it # 20:19 rhiaro tantek: there will always be challenging things in the future # 20:20 AnnB you're such a good scribe! # 20:20 rhiaro sandro: some candidates might address that better than others AdamB joined the channel
# 20:20 rhiaro cwebber2: might also be beyond user story demonstration # 20:20 rhiaro ... lots of prep work done for evaluating different existing services # 20:20 AnnB note that I thought pump.io and microformats communities were allied ... my misunderstanding # 20:20 rhiaro ... demos today were useful, but is it worth considering more discussion along tactical requirements? Especially as we did all that work gathering that # 20:21 aaronpk but they are totally incompatible implementations Arnaud joined the channel
# 20:21 elf-pavlik AnnB do you see if laptop with videoconf has some external devices connected (black device with black cable) # 20:21 Zakim sees fjh, eprodrom, harry on the speaker queue # 20:21 Zakim sees fjh, eprodrom, harry, Arnaud on the speaker queue # 20:21 Zakim sees fjh, eprodrom, harry, Arnaud on the speaker queue # 20:21 Zakim sees fjh, eprodrom, harry, Arnaud on the speaker queue # 20:21 Zakim sees eprodrom, harry, Arnaud on the speaker queue # 20:21 AnnB elf, let me check # 20:21 rhiaro fjh: In annotation wg we're just starting work on protocol work, and looking at LDP as a basis for that # 20:21 rhiaro ... sounds like you're considering not using that # 20:22 rhiaro tantek: there has been a lot of running code demonstrated that's reached a certain level of advancement # 20:22 ben_thatmustbeme elf-pavlik, sorry i thought that was your video, didn't realize it wasn't even using that # 20:22 rhiaro sandro: this process of 'you had to demo today' was never agreed upon # 20:23 rhiaro fjh: just saying that for our group we want to make our choice compatible, so we want to know about LDP # 20:23 AdamB gave up on the local wifi, using my own hotspot # 20:23 Zakim sees eprodrom, harry, Arnaud, mattl on the speaker queue # 20:23 harry So my take on things is this: It seemed micropub clearly had most of the work, and the pump.io had a large amount of maturity, and LDP had one video and not a live demo. # 20:23 rhiaro tantek: we've been asking for demos for months, that's not new # 20:24 Zakim sees harry, Arnaud, mattl on the speaker queue # 20:24 harry So my feeling is we could start with micropub and look for compatibility with pump.io and LDP # 20:24 rhiaro eprodrom: if the proposal was to consider three instead of two, any objects? # 20:25 Zakim sees harry, Arnaud, mattl on the speaker queue # 20:25 Zakim sees harry, Arnaud, mattl, sandro on the speaker queue # 20:26 Zakim sees Arnaud, mattl, sandro on the speaker queue # 20:26 rhiaro tantek: we've seen sandro's LDP demo, so that's higher # 20:26 mattl eprodrom: we need that pump/ostatus bridge thing. # 20:26 rhiaro ... we shouldn't totally shut the door, but we do have to focus # 20:26 rhiaro ... it might hold up the working group indefinitely # 20:26 eprodrom Also to get Social API and new Federation protocol into GNU Social # 20:26 fjh The idiom "Crossing the Rubicon" means to pass a point of no return, and refers to Julius Caesar's army's crossing of the river in 49 BC, which was considered an act of insurrection. # 20:27 rhiaro ... the narrowing the scope in a rough manor (not totally facist) to the top three candidates, makes sense # 20:27 rhiaro ... We start looking at micropub very deeply as a baseline # 20:27 Zakim sees Arnaud, sandro, cwebber2 on the speaker queue # 20:27 Zakim sees Arnaud, sandro, cwebber2, eprodrom on the speaker queue # 20:27 rhiaro ... and seeing what the venn diagram is between the other two proposals which have implementatoins, pump.io and LDP # 20:27 mattl eprodrom: i suggest doing the work in GNU FM first, which needs some minimal AS support anyway # 20:27 rhiaro ... What users stories can micropub not do that the others do? etc # 20:27 rhiaro ... to try to keep us on track and make progress # 20:28 rhiaro tantek: so harry is supporting narrowing down to three, plus using micropub as a basis # 20:28 rhiaro harry: focus on: what is the difference between LDP, micropub and pump.io # 20:28 Zakim sees Arnaud, sandro, cwebber2, eprodrom on the speaker queue # 20:28 Zakim sees Arnaud, sandro, cwebber2, eprodrom, fjh on the speaker queue # 20:28 Zakim sees sandro, cwebber2, eprodrom, fjh on the speaker queue # 20:29 rhiaro Arnaud: I'm chair of LDP working group, but not especially defending LDP. But I think some of this is unfair. We at least need to give people more heads up and a deadline # 20:29 rhiaro ... it's inappropriate for us right now to make a narrowing decision # 20:29 rhiaro ... I don't care about Hydra, but I've seen mentons of it. nobody said if you don't demo today you won't be in the running any more # 20:29 Zakim sees sandro, cwebber2, eprodrom, fjh on the speaker queue # 20:29 Zakim sees cwebber2, eprodrom, fjh on the speaker queue # 20:30 rhiaro sandro: wondering is it your expectation that we'll pick one and bless it and go forward with it? Or we'll do some kind of hybridisation? # 20:30 rhiaro ... If today we said actually pump.io was the best, but actually it's a starting point and we'll go ahead and modifying it # 20:30 rhiaro ... Are we keeping something existing, or modifying? # 20:31 rhiaro tantek: Personally I think running code has a very strong voice. So what is a good development strategy? # 20:31 rhiaro ... If we look at the studies of existing silos, it's clear that the way they expanded was just add a new API for every single thing # 20:31 rhiaro ... And they have hundreds of api calls andit's huge # 20:31 rhiaro ... seems like an antipattern that we should avoid # 20:31 rhiaro tantek: that has some traction, but devs do complain # 20:32 rhiaro ... how are we going to make that kind of decision? # 20:32 rhiaro ... does a user story make requirements that affect the plumbing? # 20:32 rhiaro ... if there was a clear leader, I would say can we expand that leader to take into account all user stories? # 20:32 rhiaro ... if we can, great, if we can't, we need to take some other approach # 20:33 Zakim sees cwebber2, eprodrom, fjh on the speaker queue # 20:33 rhiaro sandro: so are the people behind each of these proposals willing to express their flexibility about ... # 20:33 aaronpk elf-pavlik: how would you suggest uploading a file such as a photo within a JSON post? # 20:33 rhiaro ... if he was willing, happy to have him as editor # 20:33 rhiaro eprodrom: yes, pump.io is waiting to implement the next thing # 20:34 rhiaro sandro: right, if we think pump.io needs these extra features # 20:34 rhiaro sandro: switching from json to form-encoded, hypothetically # 20:34 rhiaro eprodrom: if we took pump.io api as baseline, then started changing it, that's pretty reasonable # 20:34 rhiaro ... there is a point at which I would take it back and reimplement it # 20:35 rhiaro sandro: would signal that with formal objection # 20:35 rhiaro tantek: haven't implemented micropub, not biased on that front # 20:35 rhiaro ... but from what I've seen, interactions between ben and aaron, and expansion of scope to handle things, seemed flexible, but aaron can speak about that # 20:36 rhiaro aaronpk: I guess priority is making it easy to implement on both sides, but especially server side # 20:36 rhiaro ... ideally a lot of different implementations of servers # 20:36 rhiaro ... from simple php script to wordpress to rails apps etc # 20:36 rhiaro ... making it very esay to handle micropub requests is important so there's a large ecosystem of clients and servers # 20:36 rhiaro ... form-encoding has been the default for a very long time, and is still the only thing that can handle file uploads # 20:37 rhiaro ... but with form-encoding the file just magically appears in the code AnnB_ joined the channel
# 20:37 Zakim sees Arnaud, fjh, harry on the speaker queue # 20:37 aaronpk oshepherd: sorry, i meant multipart. in practice they are handled seamlessly by most frameworks # 20:37 rhiaro eprodrom: I think there would be congnitive dissonance in the group by saying we're releasing a json format that is good for every situation, except this particular situation # 20:37 Zakim sees Arnaud, fjh, harry on the speaker queue # 20:37 rhiaro ... we would have to asnwer to that or fudge over # 20:38 rhiaro ... there are outstanding issues on AS around that as well # 20:38 rhiaro ... it's a working draft, anyone can propose another working draft if they have a better idea than AS # 20:38 oshepherd aaronpk: If the framework can do file uploads, it has MIME support somewhere, so JSON-in-Multipart shouldn't be impossible # 20:38 aaronpk oshepherd: I didn't say it was impossible to upload files with JSON, just much harder # 20:39 rhiaro tantek: two proposals - narrow to two candidates and develop/incorporate. OR consider 3 candidates with a similar process, with optional amendment to use micropub as a base # 20:39 rhiaro fjh: I'm still thinking about how this works with annotations # 20:39 rhiaro ... thanks aaron for explanation about why micropub is done this way # 20:39 rhiaro ... Suspect code isn't tremendously different on the server side, maybe I'm wrong # 20:39 bblfish have the two or 3 initial platforms been settled on? # 20:40 rhiaro ... From a process point of view, when we talk about it, how would we integrate/extend it? # 20:40 rhiaro tantek: don't think there's a question about ability to extend # 20:40 rhiaro fjh: not clear about how you would extend micropub # 20:40 Arnaud bblfish: tantek suggested we narrow down to micropub and pump.io # 20:41 rhiaro fjh: Other question - is this a standard, what is w3c process? # 20:41 harry However, we do have to admit it that the other specs had better demos and running code. # 20:41 rhiaro tantek: anything group considers can be along spectrum of completely belonging to w3c, to referencing other specs # 20:41 rhiaro ... we have the option of saying 'this is a great building block, lets not mess with community, we can build on it' # 20:42 rhiaro ... just like annotations is looking at adding to social spec # 20:42 rhiaro tantek: licenses for pump.io and micropub are compatible # 20:42 bblfish ah so we might go through another 3 years to just get back to where LDP got # 20:42 rhiaro ... open issue that harry is looking into to do with normative references # 20:42 harry insofar as its about a strong as the IPR in groups like IETF Guest joined the channel
# 20:42 rhiaro fjh: if we want to figure out how we can use micropub as an example, we should be looking at h types? # 20:43 harry although the non-assert is weaker than W3C patent policy # 20:43 rhiaro aaronpk: figuring out if h-entry as a base makes sense, or if there is a completely different post type that would make sense # 20:43 rhiaro harry: I liked the way we were going early where we said micropub really works, can we really use form-encoding for everything # 20:44 cwebber2 I guess I replied to something but didn't realize that was me being acked on the queue # 20:44 rhiaro ... so we can use micropub as a baseline, and look at what's missing and how pump.io and LDP do it differently # 20:44 rhiaro ... I do think we could make progress this way # 20:44 tantek bblfish, as far as demos go, LDP is 3 years behind pump.io and micropub # 20:44 rhiaro ... Looking at these is plumbing. Want to get candidates down so we can look into plumbing in a more structured way # 20:44 rhiaro aaronpk: Not blatently going to say no to everything else, just haven't heard a good enough argument to support json for micropub # 20:44 tantek code is proof. permalinks are proof. belief is irrelevant. # 20:45 harry So I would like to move to discussions re url form encoding vs. things like url templates # 20:45 rhiaro ... so far the only arguments I've heard are 'everybody likes json' and 'what about nested objects' # 20:45 tantek demos are listed in the agenda as they occured # 20:45 Zakim sees cwebber, bblfish on the speaker queue # 20:45 harry Unforunately, despite being in the same building, we did not get a demo of crosscloud # 20:45 harry but sandro notes he showed it at IndieWeb a year ago # 20:45 rhiaro cwebber2: I don't really know what I expected from the API and federation conversation, but straigh up blessing a previous work was not really what I expected # 20:46 rhiaro ... Expected that we would be rewriting our federation stuff with what came out, but expected we would be building on AS2.0 # 20:46 rhiaro ... Why not say that what we're working on as a basis is what hte group has already been working on, ie AS2.0 # 20:46 rhiaro ... whereas people have implemented micropub, pump.io and LDP # 20:47 rhiaro cwebber2: When we start talking about looking at messaging component, if you actually look at building tests for that, it already looks like an API # 20:47 elf-pavlik I don't use AS2.0 because for now I still prefer to use schema.org for 'social syntax' # 20:47 rhiaro ... Seems like we're already getting towards what an API might start to look like using that as a basis # 20:48 rhiaro harry: it would be odd if vocab and API were very different things # 20:48 rhiaro ... We we started, we were originally looking at a simple CRUD based API from open social who never showed up to wg meetings, and also looking at LDP which is now more mature # 20:48 rhiaro ... we did not look too closely at micropub; seemed like micropub is solving a harder issue, but they are all in the same space # 20:49 rhiaro tantek: regarding history of AS, when we started wg, AS was the most mature thing # 20:49 rhiaro ... so it made sense to produce a draft of that in w3c space # 20:49 rhiaro ... Weird combination of most mature thing, but community has about died # 20:49 rhiaro ... So having it adopted by the wg, james made very quick progress towards making it a draft # 20:49 rhiaro ... but still a working draft, so every single thing is still changeable # 20:50 rhiaro ... Trying to narrow things down nad make decisions, but still plenty of flexibility there # 20:50 rhiaro ... One form of input for that is social APIs or federation # 20:50 rhiaro ... If we find that in the process of doing those other chartered deliverables, we need to fix the social syntax, then we can! # 20:50 rhiaro ... If there's a problem with one, let's fix it # 20:51 rhiaro ... What needs to be done there is.. I'm not that convinced about tools... We've already built demos, we know they can work # 20:51 rhiaro ... So I take into account architectural elements # 20:52 rhiaro ... Now I think the thing for me is to see all of the other work done, if one maps it to LDP, what's missing and what's not working? Is something broken/missing in LDP? # 20:52 tantek but no one appears to be actually using it live on the web (LDP based tools) day to day? why? is there something that is broken in LDP? # 20:52 rhiaro ... that's just something we haven't thought about # 20:52 rhiaro bblfish: have to take work on LDP into consideration. Has tests and implementations # 20:52 tantek elf-pavlik: but we don't have "3 years" of LDP # 20:53 tantek that's the problem - it's alwasy a "I'm building it!" never a "I built it, here is my site, here are my permalinks" # 20:53 rhiaro ... To help narrow down on what architectural differences are. Perhaps they're not that big # 20:53 rhiaro ... ten years ago, people making arguments about rdf being complicated # 20:54 tantek q+ to follow-up to reply to facebook taking over blogging # 20:54 rhiaro ... my arguement is that these problems are there because that's the only way you can build a distributed social web # 20:54 rhiaro ... we have to try these things out, or we'll be another ten years # 20:54 tantek q+ to also note how you can build a distributed social web eprodrom joined the channel
# 20:55 rhiaro ... just because w3c standardises something doesn't mean it's a success # 20:55 rhiaro ... grddl had no uptake, was a waste of resources # 20:55 rhiaro ... it is possible to standardise things that have no uptake # 20:55 Zakim sees tantek, eprodrom on the speaker queue # 20:55 rhiaro ... our job is to create things that have wide interop # 20:55 elf-pavlik bblfish, can you record demo of *exact* implementaion of some user stories using LDP ? # 20:55 Zakim sees tantek, eprodrom, tilgovi on the speaker queue # 20:55 rhiaro ... so we should try to get consensus on narrowing scope by the end of today # 20:56 rhiaro ... plus give people one extra month to suggest candidates and defend them # 20:56 rhiaro ... but for structuring conversaion for tomorrow, we should focus on live working candidates # 20:56 bblfish yes, it would probably take some elements of all the other apis, and just generlaise the data structure very lightly # 20:56 Zakim sees tantek, eprodrom on the speaker queue # 20:57 Zakim sees tantek, eprodrom, tilgovi on the speaker queue # 20:57 rhiaro sandro: taking micropub as strawman starting point / reference # 20:57 rhiaro ... if it doesn't do something, can we look at other candidates for solutions? # 20:57 rhiaro harry: don't want to get lost in things that don't have implementations # 20:57 rhiaro ... Let's start with three strawmen with implementations, but we dont' have to be strict # 20:57 rhiaro ... no matter what we choose it's going to be subject to modificatoin # 20:58 harry PROPOSAL: We focus on the following three micropub, pump.io and LDP as a starting point, and others have 4 weeks to show demos or running code. # 20:58 harry note the term 'focus' does not mean we totally exclude everyone else forever # 20:58 harry as new things can happen and there may be problems *none* of these candidates have solved. # 20:58 Loqi I added a countdown for 4/14 1:58pm (#5655) # 20:58 rhiaro fjh: 'others' means other approaches besides those three? # 20:59 bblfish +1 for LDP ( I don't have any restrictions about the others that went missing ) # 20:59 rhiaro eprodrom: do we have a group of people willing ot put together LDP proposal # 20:59 elf-pavlik I will focus on Hydra+LDF but possibly will also use LDP in that mix # 20:59 rhiaro randall: considering annotations also looking at that, would like to help # 21:00 rhiaro sandro: might be difficult to get consensus within ldp # 21:00 cwebber2 sandro: btw do you know if we could invite another mediagoblin person to dinner? # 21:00 rhiaro sandro: why not just say now we're going to focus on these three, but if something else comes forward we will consider them # 21:00 harry Just to be clear, the "others" are BuddyCloud (XMPP), SocketHub, remoteStorage, LDF, Hydra, ActivityPump, Apache UserGrid API, and OpenSocial API # 21:01 rhiaro tantek: we have a page for more, but we went through a process of getting core users stories # 21:01 harry PROPOSAL: We focus on the following three micropub, pump.io and LDP as a starting point, and others will be considered if they have running code and demos that match the user story # 21:01 Zakim tantek, if you meant to query the queue, please say 'q?'; if you meant to replace the queue, please say 'queue= ...' # 21:02 tantek q+ to follow-up to reply to facebook taking over blogging and to also note how you can build a distributed social web # 21:02 Zakim sees bblfish, tantek on the speaker queue # 21:03 harry RESOLVED: We focus on the following three micropub, pump.io and LDP as a starting point, and others will be considered if they have running code and demos that match the user stories # 21:03 rhiaro tantek: we're past the point of people throwing up ideas, they need to come forward with code # 21:04 harry PROPOSAL: Is that we start with a strawman, that one being micropub, and then we systematically compare it to pump.io and LDP in terms of user stories. # 21:04 rhiaro ... We start with a strawman, being micropub, and we systematically compare it to pump.io and LDP in terms of user stories # 21:05 rhiaro ... I hate user stories as a way to avoid plumbing discussion # 21:05 rhiaro ... at some point we have to encounter plumbing discussion # 21:05 rhiaro tantek: not used to avoid it, used to focus it # 21:05 rhiaro ... we love to talk about plumbing, but it's so often disconnected from users and what features matter # 21:05 rhiaro ... to make sure you're working on something people care about # 21:05 rhiaro eprodrom: we're not making software for end users, ew're making a standard for devs to use to make software for their users # 21:06 Zakim sees tantek, harry, AnnB_ on the speaker queue # 21:06 rhiaro harry: lines are getting blurrier and blurrier # 21:06 rhiaro ... just want a way forward by the end of the day # 21:06 rhiaro tantek: seems to be a way forward that people can live with # 21:07 rhiaro Tsyesika: what are the grounds for picking micropub over ...? # 21:07 rhiaro harry: we clearly have, at this moment, more interop, more demos # 21:08 rhiaro ... you did a lot of work, don't want to say... but count the number of bits of running code # 21:08 rhiaro Tsyesika: we only demoed the code we wrote, there are like ten clients and loads of users as well # 21:08 rhiaro harry: I'd be equally happy with pump.io as a strawman # 21:09 rhiaro sandro: other reason to pick one could be that documentation is more structured # 21:09 rhiaro ... going through it in order might be easier # 21:09 rhiaro tantek: there are 10+ pump.io clients? there are 12+ micropub clients # 21:09 aaronpk not to mention Known and Wordpress supporting micropub which means lots of users # 21:09 harry so its kinda apples and orange, I would just like to have a way to start # 21:09 rhiaro Arnaud: Another way to select which we use as reference: how many use cases does each address? # 21:10 harry Well, we had more demos of people that *showed up* to micropubs # 21:10 rhiaro ... than 'i think there are more implementations' # 21:10 Zakim sees tantek, AnnB_, harry on the speaker queue # 21:10 rhiaro Arnaud: maybe implementors already have an idea # 21:10 Zakim sees tantek, AnnB_, harry, tilgovi on the speaker queue # 21:10 wilkie we don't want pump.io to fight micropub in some arena. we want proper interop. # 21:10 rhiaro tantek: we saw about equal user stories for micropub and pump.io # 21:10 rhiaro Tsyesika: we didn't demonstrate all. We can demo more # 21:10 tilgovi wilkie: !!! +1 !!! # 21:11 wilkie so strange to see fighting over who is the strawman... haha # 21:11 rhiaro AdamB: Arnaud's point is a good way to make a decisions # 21:11 Zakim sees tantek, AnnB_, harry on the speaker queue # 21:11 AdamB s/make a/make a data driven/ # 21:11 rhiaro tantek: this is why we did these user stories, and narrowing them down to what's important # 21:11 rhiaro Arnaud: before we select one candidate vs another, we should agree on selection criteria # 21:12 rhiaro tantek: we're talking about narrowing down not candidate # 21:12 rhiaro ... is kind of arbitrary decision, not really attached # 21:12 rhiaro ... therefore we could say we will do a three way comparison between each of these three candidates # 21:14 harry PROPOSAL: We will compare all three candidates based on (at least but not limited to) the 7 user stories we have clear consensus on. # 21:14 harry Note that we can discuss the user stories and consensus tomorrow morning # 21:15 rhiaro ... Really liked how james did comparison in AS with syntaxes. is this what we're talking about? # 21:15 harry Anyways, no - we don't expect you to match all 7 user stories # 21:15 rhiaro tantek: we have working demos for some user stories # 21:15 harry but we will look at how each candidate can or do match it. # 21:15 tantek RESOLVED: We will compare all three candidates based on (at least but not limited to) the 7 user stories we have clear consensus on. # 21:15 Zakim tantek, you wanted to follow-up to reply to facebook taking over blogging and to also note how you can build a distributed social web # 21:16 cwebber2 rhiaro: sounds like you've got lots of karma to spend # 21:16 rhiaro tantek: henry says he has been working on LDP for three years and it's a standard # 21:16 harry I'm close to 10 years but not there yet :) # 21:16 rhiaro ... how many people have been working with w3c standards for more than 10 years? # 21:16 rhiaro ... Some history: hundreds of man years of everything being based on xml # 21:17 rhiaro ... You had to build everything in terms of xml standards, otherwise why are you doing anything? # 21:17 AnnB_ been here for ~15, but never in a WG # 21:17 rhiaro ... Very little of that is actually in use on the web today # 21:17 AnnB_ ... I do know this history though # 21:17 rhiaro ... Just because years have been spent on something doesn't mean it works. Years != justification for using it # 21:17 sandro +1 Tantek "years put into a standard" isn't a quality metric Guest joined the channel
# 21:17 rhiaro timbl: Absolutely right. But what happened with XML is people decided developres wanted trees # 21:18 rhiaro ... What they looked at were things like SGML, if you go back before XML, the way people moved docs aroudn was SGML standard, very large but very weird # 21:18 rhiaro ... But people liked the fact it had these simple tags # 21:18 sandro KevinMarks, it's not that. It's about measuring quality, not sunk cost. eprodrom_ joined the channel
# 21:18 harry people forget XML is a massive simplification :) # 21:18 rhiaro ... So XML move was to take SGML standard and make something small # 21:18 eprodrom_ This might be a good conversation to have over green beers # 21:18 AnnB_ Boeing uses XML extensively for some things # 21:18 rhiaro ... Lots of companies started developing XML stacks # 21:18 harry eventually all coding will be simplified to a giant single string (hint hint form encoding) Arnaud joined the channel
# 21:19 rhiaro ... There's still a huge amount of software running XML because it's so much ismpler than what was before # 21:19 rhiaro ... On top of that, a whole bunch of people deciding we should do web services with xml # 21:19 wilkie some of us often point out that we want things that are built and used by many people before cosidering them here as serious, and yet we are using XML as the counter-example atm # 21:19 rhiaro ... Web Services even better example of standard that wasn't best # 21:20 rhiaro ... Then in between, people decided xml is a tree format so is boring as a data format, so worked on RDF data model # 21:20 rhiaro ... many people will tell you about benefits of RDF # 21:21 rhiaro ... since then, the graph model of rdf did not completely take of, and people have a craving for trees again # 21:21 rhiaro ... therefore reasonable standardisation of json, because it matches object model of js # 21:21 rhiaro ... when you do things with json you end up with the same issues as xml. # 21:22 harry I think that's a great summary of the W3C's history in a nutshell # 21:22 rhiaro ... also the magic thing about json is it's great integration with js # 21:22 rhiaro AnnB: we use xml extensively, huge in industry # 21:22 rhiaro sandro: more xml jobs than java jobs on linked in # 21:22 bret is available jobs in technology a quality metric? # 21:22 rhiaro harry: advantages and disadvantages of xml, json, rdf is a great dinner conversation.. # 21:22 rhiaro ... various people have said this working group will fail because we'll be unable to bring communities together melvster joined the channel
# 21:23 AnnB_ Harry: we need to seek the 'sweet' spots between all these technologies; find the spots where we can agree # 21:23 bret the working group needs to go a team building backpacking trip ;) # 21:23 AnnB_ waves at melvster # 21:23 rhiaro ... so as long as we are all flexible we will be able to bring a standard # 21:24 AnnB_ the best way to get to know folks ... camping # 21:24 mattl eprodrom_: see if you can implement wrapping this event up # 21:24 rhiaro ... this is why we launched now, as opposed to earlier, because we actually have vibrant communities of coders who are getting stuff working # 21:24 eprodrom_ mattl: That might be beyond even my abilities # 21:24 AnnB_ good thought, Tantek, but not everyone is an implementer .. nor is going to be # 21:25 Arnaud for the record, the assertion that XML is minimally used is ignoring the reality of the enterprise # 21:25 rhiaro tantek: that comment about failure is from major social networks # 21:25 rhiaro ... reality is, how do you build a distributed social web timbl joined the channel
# 21:25 Arnaud whether this is a good thing or not, XML is widely used in the enterprise # 21:26 AnnB_ and I applaud you guys, KevinMarks! # 21:26 aaronpk likley as a result of enterprise being generally slower than consumer web, so is still riding the original xml wave # 21:26 bret AnnB_: have you tried pump.io or known? fjh joined the channel
# 21:26 rhiaro a bunch of my scribing was lost because my client crashed without telling me :( # 21:26 oshepherd XML expresses a lot of things well that JSON does not, but a lot of things get shoved into XML which are entirely inappropriate. # 21:26 rhiaro so the last like ten minutes, you'll have to remember # 21:27 bret audio from the phone was better, as big as a hassle zakim can be # 21:28 aaronpk the audio just now was the built-in mic on the laptop, not the real camera # 21:28 bret aaronpk: are you meeting again tomorrow? # 21:35 melvster how was the meeting today, any highlights (for those following remotely)? # 21:36 bret i was only there for part so I cant really highlight # 21:37 Arnaud we agreed to narrow down the APIs to micropub, pump.io, and LDP and compare them # 21:38 Arnaud we have a couple of people working on starting a test suite for activitystreams # 21:39 melvster Arnaud: great, test suites are always good, I particularly like the interop test suite for LDP # 21:40 melvster Arnaud: having spent the last year building social apps for LDP I am tremendously impressed, I think it would be a good fit for those that want to be both social and distributed # 21:41 Arnaud ok, so, if you care you should consider demo'ing anything you've done # 21:42 Arnaud just because micropub and pump.io have code and demos # 21:42 Arnaud tantek actually proposed to eliminate LDP from the race, but it got pushed back # 21:43 Arnaud screen grab looks promising but won't be convincing tantek unless you can actually demo it timbl joined the channel
# 21:43 bret melvster: LDP definitely needs some live real world demos. show it if you got it # 21:44 melvster it only became REC last week ... you cant really expect too many demos # 21:44 Arnaud the opportunity is there, people who care need to do the footwork # 21:44 wilkie all of the ideas and innovation we *haven't* seen before are in those things we don't readily use/immediately understand. glad to see LDP here. # 21:45 melvster LDP is in its first month as a W3C REC ... I think demos will be coming throughout the year, I certainly intend to build lots of things with it # 21:48 bret yeah np, lot of people at this f2f, so definately reach out and ask for clarification # 21:50 melvster bret: thanks, just following from a distance, in case anything interesting can be reused # 21:52 Guest +1 for AS2 progress in the last 6 months in this WG SimonTennant1 joined the channel
Arnaud joined the channel
bengo joined the channel
# 22:01 bret whats the relationship between ldp and rdf? # 22:06 melvster bret: RDF is used to describe container / contained relationship in the file system, RDF is another name for linked data # 22:06 melvster minutes look good, LDP is a great base for building *distributed* and *non distributed* apps, so I think it offers a lot to the group, I like bblfish idea about using FOAF for profiles, that almost made it into the ostatus spec but not quite, I think that would have been a big plus for scalability if so, dont agree with harry on ambiguity, sometimes the data / document distinction doesnt matter sometimes it does, it depends on the use case, love tantek's p # 22:06 melvster assion, but sometimes that passion can be exclusive other technologies, may limit wider participation elf-pavlik joined the channel
# 22:12 melvster cool read all the minutes, a bit like following the ryder cup on wikipedia :) # 22:16 melvster what has interested me for the last decade on the social web are people, friends and connections ... I think the minutes are underweight those topics, and overweight API discussion, I would hope the latter could be build on the foundation of the former, bblfish's comment about using FOAF profiles was relevant there, evan actually implemented a pretty decent foaf in status.net too, I still link to one from my homepage ... some good work on using URLs to nam # 22:16 melvster e things, but Id really like to see use of URLs to CONNECT things ... much how the web of documents grew virally by using hyperlinks # 22:27 melvster actually you can browse through to facebook too since they have been the first to implement some linked data # 22:41 bret a ton of those links appear to be dead # 22:54 trackbot action-44 -- Pavlik elf to Collection - compare AS2 design with LDP, Hydra, Schema.org etc. -- due 2015-03-17 -- OPEN # 22:55 Loqi @danbri :: @elfpavlik @jasnell @HydraCG @LDPWG @SocialWebWG @kidehen @manusporny I'll tidy and send later in the week # 22:57 elf-pavlik sandro, aaronpk can we discouss collections tomrrow? ldp:Container looks very useful IMO more than as:Collection # 22:57 melvster elf-pavlik: i generally enjoy helping people, but Im not a member of this group, Id rather focus on coding, than debating standards at this point, just following from a distance, if someone asks in one of the other groups im a member of, eg the read write web group, id be happy to do what I can to assist # 22:58 elf-pavlik melvster, can you point me to the open source code using ldp:Container? # 22:59 elf-pavlik i mostly want to see use of inverse properties ldp:hasMemberRelation # 23:00 melvster well you have neither, but you need a webid first, i think currently supported is webid + tls but auth is pluggable, in the past has been supported openid, oauth, gmail, yahoo, aol, persona iirc bblfish joined the channel
bengo joined the channel
# 23:02 melvster elf-pavlik: yes that's what I did too, took me a few minutes bblfish_ joined the channel
# 23:07 elf-pavlik we have issue for that... and blank nodes handling may need some tweeking :S timbl and bengo joined the channel
# 23:20 Zakim excuses himself; his presence no longer seems to be needed the_frey and shepazu joined the channel
elf-pavlik good moring #social
Loqi Pelf made 1 edit to [[Socialwg/2015-03-17]] https://www.w3.org/wiki/index.php?diff=82905&oldid=82903
ben_thatmustbeme We will have a room that has the most ridiculous layout
RRSAgent logging to http://www.w3.org/2015/03/17-social-irc
AdamB scribe: AdamB
bigbluehat waves
harry See the great work done here for how we are simplifying W3C process: https://www.w3.org/community/w3process/
tantek Evan, from this spec? http://jasnell.github.io/w3c-socialwg-activitystreams/activitystreams2.html
trackbot Created ISSUE-22 - Need to get data on implementor interest on specific features of the as spec. Please complete additional details at <http://www.w3.org/Social/track/issues/22/edit>.
fjh jasnell what does http://asgh.mybluemix.net/ lead to, is it a web page?
jasnell_ Considering the test suite: One possible approach forward... Define test cases as Stories. Each Story is about a series of Events that have occurred. An implementation ought to have a limited number of ways of encoding that information as output. The output ought to be able to be fed into a test tool. The tester then asks a specific set of questions against each input for the test. For example: Story: "James posted a blog post at 1:23pm pacifi
Zakim elf-pavlik, you wanted to discuss mapping *vocabularies* mf <-> AS2.0 https://www.w3.org/wiki/Activity_Streams/Microformats_Mapping
cwebber2 I am fine with having http://username.foo.example/ in the spec, but explicitly us making the call of deprecating http://foo.example/user/username/, I am -1 on
Arnaud PROPOSED: change examples to have URLs a la http://username.foo.example/ instead of URNs
KevinMarks should I phone in then, or await talky?
aaronpk talky is online at https://beta.talky.io/socialweb
bret aaronpk: I'm still looking for a solid counter argument to http://www.shirky.com/writings/herecomeseverybody/semantic_syllogism.html with regards to inferencing and RDF syllogisms
akuckartz No stream at https://talky.io/socialwg ?!
mattl Android photo app -- https://github.com/creativecommons/list/tree/master/app/TheListApp