Hi, I saw there is a maximum of 64 chars for `obje...
# talk-ketos
steep-zebra-47836
06/15/2023, 12:45 PMHi, I saw there is a maximum of 64 chars for
objecta
aloof-oxygen-62624
06/15/2023, 3:23 PMWhat I do, though I don't know if Ory also advocates for this, is never store actual data in object names.
Instead, I purely use UUIDs which then map to actual data in a separate database. Makes things like renaming easier as well
aloof-oxygen-62624
06/15/2023, 3:24 PMAuth0's OpenFGA, which is quite a similar product, explicitly recommends doing things that way
s
steep-zebra-47836
06/15/2023, 4:54 PMThx for your input. Yes, this is certainly possible. But we have logical IDs from 3rd-party systems which are not UUIDs. Mapping those would create additional complexity. Also we are using those IDs to shard the data in our distributed system (this would be more difficult with UUIDs as well as the are random).
a
aloof-oxygen-62624
06/15/2023, 4:56 PMAnd its those IDs that exceed the 64 chars?
aloof-oxygen-62624
06/15/2023, 4:58 PMIf so, then i dont know. There are ID generation algorithms that work well in distributed systems as they include things like node ID, but those would then still require some kind of mapping...
s
steep-zebra-47836
06/15/2023, 5:00 PMyes, in some cases the 64 chars are exceeded. Do you have a concrete algorithm in mind?
a
aloof-oxygen-62624
06/15/2023, 5:00 PMSnowflakes https://en.wikipedia.org/wiki/Snowflake_ID
aloof-oxygen-62624
06/15/2023, 5:00 PMThose have 10bits of machine ID which are unique per host
aloof-oxygen-62624
06/15/2023, 5:01 PMAnd its obviously possible to move around the number of bits allocated to each segment depending on the specific needs of the application
s
steep-zebra-47836
06/15/2023, 5:02 PMthx a lot, I will check it out
steep-zebra-47836
06/15/2023, 5:04 PMI also found this issue https://github.com/ory/keto/issues/430 where ory is recommending SHA512 hashing. This would work but fail when I need to check all permissions of a user I guess.
s
steep-lamp-91158
06/16/2023, 12:07 PM
where did you see the limit? there should be no limit anymore, but you have to live with the payload size that you use
steep-lamp-91158
06/16/2023, 12:08 PM
so if requests take longer because of huge payloads, that's only fixable by reducing the payload size
s
steep-zebra-47836
06/16/2023, 12:10 PMHi Patrik, thx for your feedback! I'm on v0.8.0-alpha.2, Did you lift the limit in a later release?
a
aloof-oxygen-62624
06/16/2023, 12:22 PMThe limit seems to have been lifted in 0.9.0
2 Views