clojure.lang.ExceptionInfo: Fail to get-value: No implementation of method: :put-key of protocol: #'datalevin.lmdb/IBuffer found for class: nil {:dbi "datalevin/meta", :k :last-modified, :k-type :attr, :v-type :long}

https://github.com/juji-io/datalevin/blob/master/src/datalevin/scan.cljc#L222

Where does rtx come from for this get-value function? Oh, the scan macro :thinking_face:

Do you get this without using flow storm?

I did not, no. I've only been able to produce with Flow Storm.

https://github.com/juji-io/datalevin/blob/master/src/datalevin/db.cljc#L278 (`:last-modified (s/last-modified store)` in print-method defmethod print-method DB`)

Assuming this is expected to work when run later by Flow Storm? Does it need to be within a transaction or something?

Db is not supposed to be printed, this is for debugging only

I think there's an implicit assumption (by Flow Storm, and likely by others) that print-method can be called on any variable at any time.

Most of the objects are not supposed to be printed by users

I don't know that that's a stance Flow Storm would agree with. I'd assume the expectation is that an opaque string with minimal information is printed instead of throwing an error.

It definitely makes sense that printing may not be useful or contain much information, but I think it should be fixed to not thrown an error when printed at the wrong time.

Admittedly Flow Storm should also be catching this error when printing a var fails. 🤷

(And seemingly does catch similar errors sometimes. :thinking_face:)

These internal objects are all stateful

Totally understand; again, I don't think they should print anything useful that contains that stateful data, but I do think they should print an opaque string. (Even just removing the print-method would use the default Clojure behavior which does this, and I think would be preferable.)

Not sure what that default behavior looks like on a DB though. 👀

The error you saw is due to the fact that it tries to print a db that has not been initialized yet,

datalevin.db.DB@9dfe7f52

Gotcha; what would it take to fix?

We can of course be defensive in the print method, you can file an issue

👍 Thanks 🙂 Will try to get a PR together

Thx

Tried bypassing (remove-method print-method datalevin.db.DB), and I'm just getting a different stacktrace from Flow Storm, so I think I'm going to punt on this issue for now. 🙂 https://clojurians.slack.com/archives/C03KZ3XT0CF/p1691286079910589

Thanks for reading anyways!