# W3C Solid Community Group: Special Topic Meeting ## Present * [Virginia Balseiro](https://virginiabalseiro.com/#me) * Hadrian Zbarcea * elf Pavlik * Frederick Gibson * [Ted Thibodeau](https://github.com/TallTed) (he/him) ([OpenLink Software](https://openlinksw.com)) * Rahul Gupta * Jeff Zucker # Topic: Sub-resource level access control ## Trinpod approach * FG: URI for each triple, use that to apply access control and say things about that triple. Hypergraph approach: we deliver layer on top and app can figure out ??? to use in UI. Multiple resources (???) for the same resource means multiple requests. * HZ: Should be in the spec. * eP: Is there something that can be proposed? Do you have public documentation with more detail on the approach? * FG: Just having a URI representing a triple? * HZ: No, how you identify it uniquely globally. * JZ: How would this relate to permission at the level of a document or graph? If I make a document private, if I make triples in that document public, making the document private is irrelevant. How can a developer know which of those things to believe? Also across different servers. * FG: Feature not bug. That's the mechanism we use to have WebID Profile private by default and only public where needed. Same as if you have a container private and make some things in it public. * JZ: From dev POV, how can I trust that if feature is available? Header? * FG: You'd behave the same way. Make a request and get back whatever you get. * JZ: On any other server if a document is private, everything is. * eP: Current session is how Solid Protocol can address access control beyond resource level. Details like how it is advertised... In the end Protocol might say it is out of scope. This is an initial conversation. Not making a decision, but exploration. Questions should be documented. * FG: We can do several things. There are things that are easy/hard to implement. * RG: Inverse case: document is public and you mark a triple as private. When making HTTP request to the document, it is different representation for different users. Not sure HTTP is designed to work that way. Representations of documents should be only differentiated by quality or media type. Not dealbreaker but design issue. * FG: Yes. We'll add to the ETag (???) ... * RG: I don't think HTTP provisions that now... * TT: HTTP absolutely allows different users to see different representation of the same resource. We have this fully implemented in ??? * eP: Tend to agree with Ted. * RG: https://www.w3.org/DesignIssues/Conneg * eP: https://linkeddatafragments.org/concept/ * eP: https://patterns.dataincubator.org/book/qualified-relation.html & https://patterns.dataincubator.org/book/nary-relation.html * eP: http://richard.cyganiak.de/blog/2006/06/an-rdf-design-pattern-inverse-property-labels/ & https://www.w3.org/TR/json-ld11/#reverse-properties * TT: https://www.w3.org/TR/sparql11-query/#propertypaths * TT: N3 `is <predicate> of` inverts `<predicate>` ----- Chat contents (select all -> copy -> paste) because a lot went there that isn't yet here... brutal editing is probably called for. Chat and Polls Chat Polls elf Pavlik elf Pavlik says: https://hackmd.io/@solid/SyT6LraFp 9:02 Hadrian Zbarcea Hadrian Zbarcea says:@elf, did you say you live in Mexico? Hadrian Zbarcea says:I have another meeting I cannot skip at the top of the hour. 9:15 Virginia Balseiro Virginia Balseiro says:I jumped into minuting but not sure if we want to go in detail or just make notes. Feel free to jump into the hackmd. 9:20 Hadrian Zbarcea Hadrian Zbarcea says:@FG +1 9:26 me says:q+ 9:32 elf Pavlik elf Pavlik says:Fred please go ahead first 9:33 me says:HTTP absolutely allows different people to see different representations of the same resource 9:33 elf Pavlik elf Pavlik says:Ted please go ahead next 9:33 me says:Consider a GIF vs a JPG vs a TEXT representation of the same resource. Similar applies. 9:36 elf Pavlik elf Pavlik says:denotes 9:37 Rahul Gupta Rahul Gupta says: https://www.w3.org/DesignIssues/Conneg Rahul Gupta says:This is the document I am referring to... 9:38 elf Pavlik elf Pavlik says: https://linkeddatafragments.org/concept/ 9:38 me says:The Web of Linked Data can actually be seen in some of the very earliest images sketching out what became the Web of Documents. 9:38 Rahul Gupta Rahul Gupta says:It must only used to negotiate between things which, while being of different content types, carry the same information. More or less, in that there may be quality degradation. A jpeg version of photo may have less quality than the PNG alternative. An RDF/XML file may not be able to express all the information in an N3 file, but it might be an important subset. When there is a subset relationship, then q values must be used to allow the best one to be selected. Rahul Gupta says:quoting from the conneg document Rahul Gupta says:Perhaps you can interpret private information as a quality degredation... but I am not sure about that! 9:41 Hadrian Zbarcea Hadrian Zbarcea says:I don't think the point is to debate what technology is best for implementing a specific feature. The point is that sparql is not a 'one size fits all' solution. 9:41 me says:This is not a scaling failure in SPARQL. This is a scaling failure of the engine and/or other tools tested. me says:Your testing process, tools, etc., would probably be a good addition to the Linked Data Benchmark Council, https://ldbcouncil.org/ me says:(Yes, SPARQL, and RDF for that matter, is not the optimum tool for all use cases. "Horses for courses.") 9:47 Rahul Gupta Rahul Gupta says:Ted, I would like to speak to you on an unrelated Solid specification matter. Where can I best reach you? Private message to you 9:53 elf Pavlik elf Pavlik says: https://patterns.dataincubator.org/book/nary-relation.html elf Pavlik says: https://patterns.dataincubator.org/book/qualified-relation.html 9:54 Hadrian Zbarcea Hadrian Zbarcea says:Need to run, get ready for my next call. 9:55 Rahul Gupta Rahul Gupta says:Bye Hadrian! 9:55 FG Frederick Gibson (TrinPod) Frederick Gibson (TrinPod) says:Bye 9:55 Hadrian Zbarcea Hadrian Zbarcea says:@FG fantastic work, much appreciated! 9:56 FG Frederick Gibson (TrinPod) Frederick Gibson (TrinPod) says:Thanks 😃 9:56 Virginia Balseiro Virginia Balseiro says: https://solid.github.io/webid-profile/ 9:58 elf Pavlik elf Pavlik says:👍 10:01 TallTed // Ted Thibodeau (he/him) (OpenLinkSw.com) says: tthibodeau@openlinksw.com is usually the best way to reach me, though I confess, my inboxes are always overflowing, and I may not get to your message for some time. Private message to Rahul Gupta 10:01 Rahul Gupta Rahul Gupta says:No worries, I am proposing what I think would be improvement to Solid containers (by minimally changing the design) which I am privately circulating to test waters. I could really benefit from your review, if that's ok? Private message to you 10:05 elf Pavlik elf Pavlik says: https://www.w3.org/TR/json-ld11/#reverse-properties 10:18 FG Frederick Gibson (TrinPod) Frederick Gibson (TrinPod) says:Rahul, yes, I think so. Please use this channel if you have any issues: https://app.gitter.im/#/room/#graphmetrix_trinpod:gitter.im 10:18 Virginia Balseiro Virginia Balseiro says:I need to drop off, sorry everyone! See you tomorrow. 10:18 FG Frederick Gibson (TrinPod) Frederick Gibson (TrinPod) says:Bye 10:18 TallTed // Ted Thibodeau (he/him) (OpenLinkSw.com) says:Sure. Don't feel like you need to wait for my response before you add a PR... Private message to Rahul Gupta 10:19 Rahul Gupta Rahul Gupta says:Thank you! Private message to you 10:19 elf Pavlik elf Pavlik says:👍 10:20 avatar elf Pavlik elf Pavlik says: https://github.com/solid/webid-profile/pull/112 elf Pavlik says: http://richard.cyganiak.de/blog/2006/06/an-rdf-design-pattern-inverse-property-labels/ elf Pavlik says:you are next Fred 10:25 me says:A quick and dirty `DESCRIBE <entity>` result just gives you `<entity> ?p ?o` unioned with `?s ?p <entity>`... 10:27 FG Frederick Gibson (TrinPod) Frederick Gibson (TrinPod) says:👍 10:27 me says:`is <predicate> of` inverts `<predicate>` 10:28 avatar elf Pavlik elf Pavlik says:👍 10:38 Type a message