yes, that's usually how it goes

that's why its the default

I quick question, I found dot is not supported in resource ID, like workflow_type.payment

Is it reserved?

that's the regex

its simply not allowed right now

I see thanks !

What does this error mean whenn checking permissions:

rpc error: code = Canceled desc = unable to read namespace config: context canceled

What SpiceDB version are you using? this should have been addressed recently

I'm using v1.13

Yeah I believe this was fixed in 1.14.0

so what do you recommend for v1.13? retry the permission check?

We recommend upgrading to 1.14 😅

😄 . I can't right now because of mysql version.

what's up with MySQL version? the caveats table size limitation?

yes

bummer. I wonder if there is anything we could do about that.

Hey everyone, I'm evaluating Authzed/SpiceDB and I'm running up against a problem, that I'm sure others have solved but I couldn't find anything anywhere. SpiceDB seems to have some very specific requirements on what characters are allowed as resource and object ids https://github.com/authzed/spicedb/blob/5009a53ac4ecb2f57ef7f6d9ed2444d166a6fa1d/pkg/tuple/tuple.go#L24-L25. My simple usecase, is a user who's id is their email address. I thought maybe base64 encode it, but forgot about '=' padding. What kind of encoding do folks suggest for resource ids to ensure compatibility?

There are base64 encoding schemes (e.g. the ones used in JWT) that drop the padding. Those should work fine.

That is amazing, I'll check that out!

Ah Base64Url!

Hey perhaps a silly question but is this note from the docs correct? > # memdb > > ## Usage Notes > > * Fully ephemeral; no data is lost the process is terminated I'd have expected the memdb to lose data when the process terminates. Although; i'd really like a persistent data store for dev/test that doesn't require an additional container 🙂

Yeah that seems wrong. Feel free to open a PR to fix it up, or we will fix it tomorrow

No worries; I can't make a PR unfortunately. Thanks for the fast response!

SQLite datastore wouldn’t be that hard

yeah, we'll probably use pg for dev and memdb for integration/unit tests for now.

That’s what we do! Well, we also have a crdb dev stack.

Did you see the

serve-testing

command?