unsure

if it doesn't, we'll do a release on Monday with it

looks not wrong...

port: 8080

?

50051 should be to 50051 and 50053 to 50053, I believe

thats just the port exposed by the load balancing service

ah, right

I'm not sure that it matters

nah

its fine if you're going through the LB

its just unexpected

I don't see, com.authzed.api package inside it , it has only come.authzed.grpcutil package 😞

you're pulling from Maven?

No checked out latest code from Java client repo

use Maven

Java client lib uses a special buf rule to pull in the protos

Sure thanks, will do this

Is there any easy way to handle that rule, working with client lib source code comes very handy.

I’ll have to ask my coworker on Monday

Thanks,I imported the repo in intellij with java 11 its not compiling

Just checked maven one has lookup subjects client

When we create a relation , we would like to associate some metadata with it. As example , one whose request the relation was created and when was it created. For auditing purpose. Is there any such thing exist or do you plan to build something like this ?

👋🏻 we've discussed it internally, but we don't have an issue opened in the repo yet. The work we did for caveats provides a nice foundation for this, but we need to figure out if we want the metadata separated from caveat context or not. Would you mind opening an issue with your use-case? 🙇🏻‍♂️

Is there a way to delete a resource, leading to deletion of all relationships into and out of that resource in a single request? e.g. A folder has contents, but an organisation has folders, so if I delete a folder, I'd like all organisation->folder and folder->contents relationships to be deleted in one go for that folder

no, we don't support delete semantics like that. Feel free to open an issue tho!

Thanks @vroldanbet . So is the answer then that the application figure it out, and, if absolute consistency is required, we lean on something like Kafka per the docs?

correct, the application should figure it out. Mind elaborating what you mean with "if absolute consistency is required"? As in using Kafka for bookkeeping the deletion process?

To use an example scenario. Users can create, for example, folders within folders. So if there's a folder hierarchy A -> B -> C -> D. A is the root folder. User 1 deletes folder B when User 2 creates a sub-folder X inside C, how would you recommend we prevent orphan relationship scenario where A -> B no longer exists, and many of the other relationships were found and deleted, but a race occurred so C -> X still exists, so direct requests for C respond with the relationships to X?

From my perspective the application is the source of truth for your business domain. It should have enough context to be able to walk the graph and issue a deletion for the whole subgraph transactionally. As for the race-y subfolder creation - don't we have the same challenge in the client application? Presumably you'd do something like "do not create this folder if the parent does not exist".

WriteRelationships

in SpiceDB expose preconditions as well so you could potentially do the same. The challenge here is that in SpiceDB there is not concept of "entities" - an entity exists if it's defined via a relationship, so you'd need to find an appropriate relation that allows you to make sure a folder exists.