rhiaroeprodrom: At the f2f we had gone through some additional requirements for the document including implementation report template, that kind of stuff, those are all now in the documents
rhiaro... And a link from the validator to ... one thing that I did with this last push is that I moved the test data into the spec document repo, it was easier to keep everything all together. May do that with the validator code too
rhiaroaaronpk: I just found a random URL in the test suite and dropped it into the validator and it gives a warning that it should have an 'id' property, so I thought that it might be good to check the test suite data actually validates
rhiaroeprodrom: with AS2, there are very few required properties. A name is almost always required, an id is usually required, so we give warnings. We have examples.. one example we have is an empty object. So we have examples that processors should be able to consume and should be able to parse, however when they're put through the validator they'll show warning because they're poor style. Does that make sense?
rhiaro... If you put an empty document through as2.rocks ... I think need to have a testable system for consumers and we need to have test data that's useful to help them test edge cases, but we also have to have those notifications in the validator for those edge cases
rhiaroeprodrom: What should your code do? Send an email, fetch a url, put a pin into a map. I'm not sure if there's a correct behaviour we can require out of these processors. What we're looking for is can you parse it, does it look right within your framework
rhiarocwebber2: I just wanted to follow up; we've discussed this before, AS2 is a vocab and the testing toosl around it don't and can't enfoce side effects. That's what AP and ActivitySub do, they're examples of real requirements of implementing that. This has come up before and we specifically went against requiring side effects to have in the tests, as this isn't specified in the vocabulary
rhiaroeprodrom: One thing we could do is are you looking for some comments on this? Like this is a place, this has all the optional attributes of a place... something like that?
rhiaroaaronpk: The other way to handle that is are there examples of documents that are known bad, so I know that if I parse this document I should see an error? That would help give me something to test
rhiarosandro: but mostly I think you have to hand inspect each one. You're supposed to look at the document, read the spec, and see if your app does whatever makes sense to do for that content. That's my expectation of what it means to be a conforming consumer. You can't automate consumer testing of AS2
rhiaro... The implementation report around consumers presumably say something like my appication understands what this vocabulary temr means and does something meaningful with it. There's no way to automatically test that given it's all human behaviours.. it turns into some html or pixels on the screen
rhiaroeprodrom: Yeah the implementation document covers all of the object classes, each property of each object class, the core and extended vocab, and asks whether you implemented that class, and for each property did you implement it, and a place for comments
cwebber-arghalso we had a heavy discussion about this in SF and after and came to the conclusions about how to go about testing and etc, I really feel like this has been move forwarded with after already being discussed to death
rhiaroeprodrom: We say in our document your application must use this, just passing through unrecognised properties doesn't count as an 'implementation'
eprodrom"For each core class your application implements, note which properties of the class it uses. Here, "implements" means that your application uses the property directly; just passing through unrecognized properties doesn't count as an "implementation"."
rhiaroeprodrom: I think so. There may be some trickiness with the fact we've moved to a different ED location. I'm not sure if the echidna token stays valid...
rhiaroben_thatmustbeme: I ran out of the meeting early last time and we resolved to get there. I don't know where to go next. I need to work out with sandro to get a token to publish
rhiaro... I looked through exit criteria for webmention. Report for GNU Social has a couple of unchecked boxes, I assume that's going to be the case for all implementations
rhiaroArnaud: Do you think that we can .. I don't think we can freely change places where the spec says you SHOULD do this.. we can't just take those out or add those
rhiaroeprodrom: the only thing that I'm getting out of.. looking at the gnu social implementation.. seems like they skipped some of the security and verification suggestions, which I think are that's up to you if you choose not to implement the security ocnsiderations, but if we see a pattern there where we see no-one implements the chekc of not accepting webmentions that are in an html comment, I wonder if that informs any decision making we do later on. Do we
rhiaro... We're suggesting some verification, validation and security parts of the implementation. We have at least one case where somoene has skipped a lot. If we see people skipping a lot is that cause for concern for us?
rhiaroeprodrom: I'm going to keep an eye open. i think that in writing most of our specs we are saying it's okay to do x because you will also do y. It's okay to include this link in your page because you're also going to be verifying the link was in this webmention. It's okay to include unverified properties in your AS2 because you're going to be checking this and this. But if you don't do the checks, i makes those other parts more dangerous
rhiaro... In security I think mine.. I don't moderate, I displya them immediately. I don't reverify. I don't have a byte limit, but I do have a time limit
rhiaro... One thing is that jessica and I have done some large refactoring of the document that's making things structurally better to reduce effort when we split into two documents
aaronpkben_thatmustbeme, i don't think it needs to explicitly say ignore links in comments since it says to explicitly look for <a> <img> and <video> and similar tags
rhiaro... And there's a PR from jessica that we're looking at merging, we just need to refactor it to the existing refactor that just happened. THerew as also some discussion we should have next week about authentication stuff
rhiaro... I supported that from a consumption standpoint, but I'm still outputting @id and @type. In a certain sense it's easier for me because I'm using json-ld tooling
aaronpkcwebber2: if someone wanted to start getting involved with and using activitypub and activitystreams today, what would you point them towards as far as resources and examples?
cwebber2aaronpk: for activitypub, the standards doc and the pump.io api page are probably the two clearest pages. As for ActivityStreams, I think the docs I wrote on ActiviPy are pretty clear http://activipy.readthedocs.io/en/latest/
aaronpk well that's odd... why am i seeing the github pages 404 page here? https://www.w3.org/respec/ref.html
RRSAgent logging to http://www.w3.org/2016/06/21-social-irc
paul present +
ben_thatmustbeme present+
KevinMarks I'm on but muted, didn't hear aaron
Arnaud RESOLVED: Approve minutes of June 6 2016 https://www.w3.org/wiki/Socialwg/2016-06-06-minutes
rhiaro eprodrom: with AS2, there are very few required properties. A name is almost always required, an id is usually required, so we give warnings. We have examples.. one example we have is an empty object. So we have examples that processors should be able to consume and should be able to parse, however when they're put through the validator they'll show warning because they're poor style. Does that make sense?
cwebber2 aaronpk: for activitypub, the standards doc and the pump.io api page are probably the two clearest pages. As for ActivityStreams, I think the docs I wrote on ActiviPy are pretty clear http://activipy.readthedocs.io/en/latest/