#social 2020-04-25
2020-04-25 UTC
# 
nightpool[m] can you send an example? I'm having a hard time picturing what you're talking about jaywink
# dansup Pixelfed is now 2 years old! https://pixelfed.writeas.com/pixelfed-v0-10-9
jussi, emilis, jussi_, jussi__, pukkamustard and emacsen joined the channel
# emacsen Hi. Is there a meeting today?
jussi, rhiaro_, rhiaro, sl007 and cwebber2 joined the channel
emacsen joined the channel
Zakim and RRSAgent joined the channel
# 
RRSAgent logging to https://www.w3.org/2020/04/25-social-irc
hellekin joined the channel
# hellekin hi there
# sl007 present+
# hellekin present+
# pukkamustard present+
# emacsen (should I be scribing here
# emacsen cwbber: I am Chris Webber. I am chair of SocialCG
# emacsen Hellekin: I am the cat herder of SocialHub
# emacsen punkamustard: I am using ActivityPub in [inaduble] project (pls fill in)
# emacsen thebaer: I am the author of WriteFreely and other ActivityPub projects as well as FreePhone
# pukkamustard I am working on a project called openEngiadina using ActivityPub for local knowledge such as Events, status informatin, local organizations and businesses etc.
# sl007 Hello, I am Sebastian, doing redaktor.me
# emacsen Chris: Next topic is content addressed vocab. I wrote a threat and made a blog about it
# emacsen Chris: Right now AP uses http uris by extensions
# emacsen Chris: A question of where should extensions go?
# emacsen Chris: Where should the URI of where a vocabulary goes and where does the activity context go
# emacsen Terms first. We spent 6 months in SocialCG looking at extensions to vocabulary in ActivityStreams.
# emacsen Maybe we were approaching things the wrong way. For example when Mastodon introduced sensitive field.
# emacsen It was pushed to put in ActivityStreams core, but what happens to older implementations
# emacsen What happens to older software, or outdated urls.
# emacsen What is governance process for putting terms in the vocabulary and what happens when do they stabalize
# emacsen This is a hard question to answer and we don't have a resolution.
matt joined the channel
# emacsen I used to be pro-namespaces, and I have become less so over time but we still need global unique identifiers
# emacsen URNs are a valid URI that can be used for this.
# emacsen We could use a hash before the identifier. What is this hash?
thebaer joined the channel
# emacsen Instead of having http before the URI, we could have the specification text and have it as the name of the term
# emacsen We could use JSON-LD contexts to map "sensitive" to something less ambigious- the hashed URN
# emacsen That's the general idea. The flip side is that with JSON-LD contexts, the contexts could go down or change.
# emacsen If it does, then the lookup or signatures could fail. If we use Datashards or other approaches, we could address these problems.
# emacsen I'd like to see what SocialCG thinks of this.
# emacsen Does anyone have any thoughts? Use q+ on IRC to queue yourself up.
# emacsen sebi: Can anybody hear me now?
# emacsen cwebber2, Yes. you're audible
# emacsen sebi: Yes
# emacsen I'm Sebastian [inaudible] I manage redaktor.me where you can discuss ActivityPub activitivy. We have a [inaudible] map where you can replace [something] with ActivityPub
# pukkamustard q+
# emacsen pukkamustard: I like the approach of using content address namesapces. There were questions
# emacsen pukkamustard: Hashing the content. This was not in the blogpost. Can you elaborate?
# emacsen cwebber2: So basically the idea is that you use some sort of CAS such as IPFS or Datashards or something like that. Something globally addressable
# emacsen if you're not familiar with CAS, then instead of a uri example/cat/photo you'd use the hash of the content
# emacsen cwebber2 I'm working with Serge on Datashards, which has more privacy than IPFS
# emacsen cwebber2 If we use a CAS system, anyone who has the content can keep the content alive
# emacsen cwebber2 That's valuable in that you can keep it alive, and that's the main appeal of JSON-LD content addressed content
# emacsen pukkamustard On the blog post you hash the individual properties, rather than the context
# emacsen cwebber2 We are solving two different things. One of them is hashing the name of the term to make a unique name for the term. If it's retrivable then we know it refers to this term
# emacsen That means that we can look up a definition.
# emacsen JSON-LD contexts by providing mapping are also important to be unique and to keep alive but it's kind of for a different reason
# emacsen It's not the same problem but has the same solution
# emacsen pukkamustard, I'd like to elaborate on the difference between JSON-LD contexts and namespaces
# sl007 q+
# emacsen pukkamustard, JSON-LD contexts can't be mapped to the namespaces
# emacsen sl007 Instead of using these describtions for the RDF namespaces, we use JSON schemas
# emacsen sl007 But that is just our solution and not an official one
# emacsen cwebber2 I think there are differet things going on here. In some RDF systems you use shorthand for URIs using namespaces. JSON-LD just uses a mapping of terms. Not all RDF systems have these kinds of mappings of a string to a specific terms, rather than to a specific namespace
# emacsen cwebber2 so one argument could be that by not using namespaces, it is less useful for systems that don't have this mapping. I forget if Turtle allows you use certain namespaces
# emacsen pukkamustard You could give namespaces to a specific term
# emacsen cwebber2 But you can't pull up a bundle of them at once. JSON-LD may be more easily equipped to deal with this solution than other RDF serializations even though all of these are RDF serializations
# pukkamustard q+
# emacsen cwebber2 I haven't thought of this much and don't know much about other RDF serializations
# emacsen pukkamustard, I've been toying around with CAS, specifically Datashards. I've been trying to find a serialization of general RDF data. This would allow any existing namespace of content addressable namespaces such as activitystreams content
# emacsen q+
# sl007 To eleborate we use https://json-schema.org at the JSON LD context URL with Content-Type application/schema+json
# emacsen cwebber2 There are two things here. Even if we stick with namespaces, we could use mutable datashards for a namespace and update the namespace.
# emacsen cwebber2 The thing I'm arguing for is not using namespaces to avoid governance issues even though other RDF systems assume you use namespaces
# emacsen cwebber2 Using CAS such as Datashards are useful for storing data in the usual case, such as when witches.town went down, it protects data in the general case
# emacsen pukkamustard, Yes.
# 
cwebber2 emacsen: one thing I would caution to any datashards implementers is that datashards has gone through multiple iterations and transformations in the last few months that have not been reflected in the documentation and git repository mainly due to the work that we're doing in response to this issue specifically
# hellekin q+ reply to emacsen
# emacsen hellekin: Serge, why don't you put this topic online so other people can help around?
# cwebber2 emacsen: the reason that this hasn't been widely documented and feedback solicited from others is that there are some core principles that up until a literally few days ago were not hammered out and it made sense to wait for something discussible, otherwise we'd end up with a lot of radically different implementations... in the next month or so I should have a functional datashards implementation that captures all of the changes, then
# emacsen cwebber2 I wanted to ask pukkamustard, I'm excited to hear that you're excited. Would you like to expand on the state of your work?
# pukkamustard q+
# emacsen pukkamustard, I'd love to. This is for the project openangidina. It's exactly the same as Goblins to make it more robust and make content more available and cryptographically verifiable.
# emacsen pukkamustard The stage is that there is an implemention in Guile. The primitives are difficult to implement, so I used libSodium to make things easier.
# emacsen pukkamustard This is something I'd like to explore. Then I've been using RDF to store the ID of the content rather than the content itself.
# emacsen cwebber2 Let's try to find some time to talk
# emacsen pukkamustard That sounds perfect
# sl007 q+
# emacsen sl007 About Datashards JS. I would then know if I could switch to libsocial. Right now we are using the Digital Bazzare stuff. It is on github by the way.
# sl007 libsodium
# sl007 what pukkamustard said
# emacsen cwebber2 Maybe there is more interest than Serge and I realized. I'd like to sync up soon on this
# emacsen q+
# sl007 +
# emacsen cwebber2, We don't need to record it but I could write a summary if there are any takeaways
# hellekin q+
# emacsen hellekin One major takeaway is that you must mention Serge is that you would have something in months. That would be good to let people know to hold their horses
# emacsen cwebber2 Moving on, end to end cryption with Redaktor news
# emacsen sl007 As I wrote in socialhub we interrupted this for the map work. What we did now is we does anyone know the standford password protocol where you get a shared AES encryption key.
# emacsen sl007 we have a shared key but it not submitted over the wire but we use an RSA key and AES key but we switched to the map thing. What I would like to do is think about how to represent a map in ActivityPub.
# emacsen sl007 About the map. Every timeline can have multiple places in every post but if we want to represent it on a map we can represent the post itself but in crisis response a user might want to say I want to attach these layers to a map, which is the classic use of attachments
# emacsen sl007, So we use a map to the layer and the map is a collection of map layers to the map, such as a topographical map where many information is stored in GeoJSON which is not activitypub
# emacsen sl007 We attach this as map layers which can be attached as map layers which can be attached to your general collection. Is there a better way?
# emacsen cwebber2 I feel like there was a lot there to see if there is a better way. I would want to see it visually/structually to see if there's a better way
# emacsen q+
# emacsen sl007 You can use the URL of ESRI or OpenStreetMap and this together makes a layer. Is attachements the correct place to put it in ActivityPub?
# emacsen sl007 It all works perfectly with Leadlet
# emacsen leaflet
# emacsen sl007 Together with the vector plugin where you can use the vector plugin where you can interpret every vector
# emacsen cwebber2 I see no one on the queue and have no mapping experience.
# emacsen cwebber2 Any responses?
# sl007 emacsen like http://leaflet-extras.github.io/leaflet-providers/preview/ but you can also combine layers
# emacsen cwebber2 Thank you Sebastian for updating us on that.
# emacsen cwebber2 We have 10 minutes left. Bear would you like to discuss c2s widgets
# hellekin thank you sl007 for the update on Redaktor.me. Cool stuff!
# emacsen bear: We discussed this at FOSDEM but no work since then. Quick update on WriteFreely is that we're looking into submissions, where people can approve or deny posts or replies for a blog format
# emacsen bear: Right now it's high level UX but it's going to work towards low level UX is worked out
# emacsen bear: Ideally in the future we'll have cross instance support for new kinds of replies and submission mangement software.
# emacsen cwebber2 Cool. I wanted to see if I understand. Are you building compostble web UIs?
# emacsen bear: Yes
# emacsen cwebber2 Very cool.
# emacsen cwebber2 Do you have any links?
# emacsen bear: Not yet. As soon as we do it will be in the form of the writefeely repository.
# emacsen cwebber2 Thank you very much!
# thebaer WriteFreely repo: https://github.com/writeas/writefreely
# hellekin thebaer: please open a topic in C2S on socialhub when you're ready!
# emacsen cwebber2 Do we have the people here who represent the next topic? Reaching out to others.
# sl007 q+
# emacsen cwebber2 Are they here? Or the Groups in AP people here?
# emacsen sl007 About Reaching out to others we have this huge huge proposal for a guide for the home page where we reach out to others
# thebaer hellekin: will do!
# emacsen cwebber2 We have the home page as a topic a few meetings ago. The last progress I remember is a few weeks ago and the proposal has been made
# hellekin ?
# emacsen cwebber2 Could you link to it on IRC?
# emacsen cwebber2 I'm pulling it up now
# emacsen cwebber2 Is the goal for this what we discussed previously to put something like this on the home page of activitypub.rocks or a link to this on activitypub.rocks but I thought the idea was to link to it as a first/next step?
# emacsen cwebber2, I'm aiming that at you Sebastian
# emacsen sl007 The progress as per the IRC link is that the guide was split one for developers and one for end users and maybe we could put this on activitypub.rocks but someone needs to make an HTML page from the post
# emacsen cwebber2 For now how about I put it on my agenda to do that tomorrow? I'll schedule that right now.
# emacsen cwebber2 Hellekin but this is the whole thread. Can we make a nice guide from the page somehow?
# emacsen cwebber2 I'm all for that
# emacsen sl007 Or socialhub But the link I posted is with all the answers with the making of the guide
# hellekin q+
# emacsen cwebber2 So you're taking this and transforiming it into a different page?
# emacsen hellekin So if we change the title the slug will change but the ID will be chaged. But it will be redirected to the right one
# emacsen hellekin, I just set a setting so no discource on this so that people can publish with their own style
# emacsen hellekin, We could keep the wiki to amend and have a separate one to publish
# emacsen cwebber2 Could someone create an issue with what I should link to from the home page since I don't know what the pages to link are.
# emacsen sl007 Yes, but the hellekin needs to be set up in its own url
# emacsen cwebber2 I will leave that to you two and the I can act on it
# emacsen cwebber2 We're three minutes till the hour. The last thing on the agenda is next meeting scheduling
# emacsen cwebber2 We used to have a formal schedule but it's dropped. We could agree to go back to the bi-weekly approach or go back to this time or monthly
# hellekin q+
# emacsen hellekin I would repeat the suggestion I posted on socialhub that we could schedule a topic in the next 15 days that it's a wiki where people could add items they want to discuss and if they're discussed on the forum in the meantime and if they're not checked the it's put on the meeting agenda or the next monthly meeting would be mandatory
# emacsen cwebber2, I hear what you're saying but I would caution that I'm not good with too much process. This is why I brought Nightpool in but he's been busy
# thebaer q+
# emacsen cwebber2 I would suggest that people can use this channel if they want to discuss topics but I would like a regularly scheduled meeting for people to know to show up
# emacsen hellekin The problem with time is the time zones
# emacsen cwebber2 Yes it is, but it's inescapable
# emacsen cwebber2, I think that I will operate best with a set time
# emacsen bear: I am okay with a monthly meeting
# emacsen cwebber2 Why don't we keep it simple and vote on keeping it the 2nd saturday of every month.
# emacsen cwebber2 I'm going to propose with but won't be the person to create the page
# emacsen +1
# hellekin +1
# sl007 +1
# emacsen +1
# thebaer +1
# pukkamustard +1
# emacsen cwebber2 Full conesnsus. The next meeting is next month. Someone not me, please make the page
# emacsen cwebber2, Let's end this one.
# emacsen Meeting ended
Zakim left the channel
# RRSAgent I have made the request to generate https://www.w3.org/2020/04/25-social-minutes.html cwebber2
Grishka joined the channel; RRSAgent left the channel
# Grishka is it too late to join?
jussi_ joined the channel
# hellekin may I ask the people coming from Matrix what is the Matrix room for this channel?
# nightpool[m] cybre.space:socialcg
# nightpool[m] er, other way around
# nightpool[m] uploaded an image: Screenshot_20200425-141328.png (70KB) <https://matrix.cybre.space/_matrix/media/v1/download/cybre.space/WPhdolCnWZpadnhVcsPqplwS>
lanodan joined the channel
# 
jaywink[m] btw nightpool this (matrix side) room is still v1, might want to consider upgrading it at some point, to v5
# jaywink[m] lots of fixes between those room versions
emacsen left the channel
# nightpool[m] what are room versions?
# jaywink[m] the matrix protocol rooms are versioned. different room version means for example different state calculation algorithm. in this case v1 is buggy while the current v5 is the version that was out when matrix protocol was given a 1.0 version
# jaywink[m] basically you just type `/upgraderoom 5` as admin. the IRC bridge might need some tweaking if it doesn't automatically survive the change. all participants will be invited to the new room
# jaywink[m] the worst thing about v1 rooms is the occasional state reset which can lead to different servers having a different view on who is in the room
# jaywink[m] miight want to ensure you run a relatively recent matrix-appservice-irc version before upgrading tho, lots of room upgrade issues were fixed I think in recent versions
# nightpool[m] got it
# nightpool[m] well, I'll take a look at some point
hellekin and emacsen joined the channel