# SolidOS Team Meeting

* Date: 08.06.2022 18:00 (UTC+1)
* Call: [https://meet.jit.si/solid-operating-system](https://meet.jit.si/solid-operating-system)
* [Previous meeting link]()
* [Next meeting link]()
* Chat: [https://gitter.im/solidos/solidos](https://gitter.im/solidos/solidos)
* Repository: [https://github.com/solidos/solidos](https://github.com/solidos/solidos)
* Meetings home: [https://solidos.solidcommunity.net/public/SolidOS%20team%20meetings/](https://solidos.solidcommunity.net/public/SolidOS%20team%20meetings/)
----
## Present attendees
* Alain, Angelo, Noel, Timea, Aad, Tim, Chase

### Scribes
* Timea

---

## Topics

### Round Table

#### Alain
* cheked on NSS the number of Pods without email info. I updated the issue with migration to CSS. https://github.com/solid/solidcommunity.net/issues/62#issuecomment-1146657049 There are more than 10k accounts.
* on CSS email is required, a unique email, there is no other way to log in.
* needs ot be an email form - you do not need to receive the email. User needs to know what they used to login.
* right apporach is to create another authentication solution for NSS. -> we can copy the CSS but without the check of email.
* How shall we go about it?
* ACTION: Timea to talk to Ruben and see what is the best approach here.
* We could code the module even outside the CSS repo - we need knowhow and time.
* Alain: we will have to maintain it.
* Angelo: best is if CSS supports the module.

#### Angelo
* working on onboarding experience. Am about to finish it next week.
* I struggeled with trusted app on NSS - needs to get in the trusted App - with the chrome extension it has a fake redirect - I had to do a workaround for that.
* browser extensions are more previleged than browser sites - still it was quite some work to implement.
* afterwards I will go back to SolidOS stuff.

#### Noel
* I started implementing how to delete recipies - offline first is not so easy - if I delete on one device and aldo on another device - how would it reconsile?
* soft delete is an option - another option is to delete in the pod directly
* I will probably the 'delete the data for real' but I have to think about it- If they are not online and they are not connected to pod - how do I show it? any better ideas?
* 1st version - will mark it as soft delete and when they connect it will get deleted for real. In next version I cn show a que of deletion. I wasnt to see if there will be any problem
* I opened a new PR in solid vocabulary - for chnage detection - Tim did that review - Thank you!
* one operation I had was 'type'. when you do an operation to that resource, it can have different types (set, unset... ) and the point that Tim raised: type of operation cannot have default. A subgraph is a valid graph. If I say that the type is remove and it has a default add it would be a contradiction. I saw it on schea.org -> defaults for properties ->
* I got a solution - instead to have operations for type property I have subproperties. I need ot see how to implement it. My lib was looking for resources of type operationand if I have to support subclasses I need to change that.
* What do you think about property defaults?

#### Timea
* Catching up with communication after vacation.
* Tests are failing on solid-logic branch. Collaborating on PR
* Will focus on move to CSS (communication with CSS) as well

#### Aad
* the solid shoping ecosystem - there will be an intern working on it and Matthias (ewingson) - smth like amazon and eetsy on Solid. If you have a name for it let me know. Emptor was a suggestion.
* Oslo event
* What we noticed in Flanders - we need more devs we can attract. There are active on courses with SPARQL - we give projects and it does work. We decided to set up a space for questions, on matrix. A space can be smth with various rooms. Go to chat.semantic.works or on Matrix on LDS (Linked Data Support)
* I fully agree with Jon and Tim -> this should be in out Solid Pods. Why it makes sense is if it talked more about Solid than LD. For now we talked LD. Future work: sync all of this with SOLID.
* How to join the chat (pick one):
- Go to chat.semantic.works; register and login; join some rooms
- Visit your Matrix server; join channel; enter #lds:chat.semantic.works

#### Tim
* adding an unknown origin to acl file does not work - I needed to give timbl.com the right user - was not easy to set up timbl.comas a trusted app.
* community type indexes - when you are part of a community, you can make a note and store it in your type index and also in teh community type index. Hopefully, wen working togetherm make a new tracker and we can commonly share it. If you go to my trackers you see mine and also the ones form teh community.

#### Chase
* last week, Noel, Angelo and I spent an over and learned more Visual Code -> compilation of code is much faster. Got a clearer view of the code, cannot live without it!

### Technical topics

#### What do you think about property defaults?
* Tim: interesting discussion abotu schema.org. They avoided to getting caught traped in logic. We have a domain and a range. You can use it but it does not mean it is a 'person'. In a particular app yes but not in ACL. You sign something and I can prove it is true.
* it is bad to design things with a default, it's a trap.
* we will probably go towards using rdfs inference as Solid developes.
* Noel: conceptually using inference is teh best - but complexity implmeneting the code and to do inference you need the whole vocabulary. If you try to use schema.org propery it does ont load the whole context 1MB, imagine if I have to download the context for all.... that is an issue.
* Tim: if you look up schema.org/person -> do you get whole?
* Noel: json-ld is 1MB - is it an implementation?
* Aad: I think it is the implementation. In practice, we reason in a particular target model, if you do it so you can do it automatically. This holds for all reasoning I think. We did inheritence reasoning and it had a large performance impact. It is not going to be feasible with OWL in all cases. downloading schema.org and it is fetching data from users... webpages, it is fine in that case. In terms of modeling they care about technology that lets them annotate not reasoning. GoodRelations is it the other way around. In Flanders they use some defaults: entities in a certain timeframe, by default it taes all properties of the default except the one that were ovverriden. It is weird in an open world assumption. It is used in the Oslo standard model.
* Tim: because you want close world in your ontology, vcard you know what is reasonable to say
* Aad: will show next time a potential solution for Noel.
* Aad: on deletion -> move data ina separate graph.
* Noel: if user deletes the data I want taht the data does not exist anywhere.
* Angelo: what about deleting the recipe resource but keeping the metadata?
* Noel: I could leave a 'tombstone'.
* Tim: when deleting - people could be sharing the recipe stack - if I worry that someone accidentely deletes smth - I hit ctrl-Z-> very valuable. IN calendar we see renaming and I want to see who did it. In CRDT you get a list of all changes to undo. One can do a review.Then you get the whole collaborative system.
* Noel: i agree and I should have it but I do dnot like system where you tell them to delete. One could habe 'delete' and 'incinerate' (like from ruby on rails). on 'deletion' it could be incinerated after 2 weeks.
* Tim: in SolidOS we should have a trash can.
* Noel: in my app i will delete it forever - with proper warning.

----
This template is based on the [W3C meeting template](https://github.com/solid/specification/blob/main/meetings/template.md)

[Code of conduct](https://github.com/solid/process/blob/main/code-of-conduct.md)