Asmir Avdicevic

Our error handling is highly technical, and cryptic in some cases even to us. Users shouldn't have to go into code to figure out what an error means. Linking to...

Got the log bellow a couple times while running tests, would like to sort out these concurrent map read/writes for stability reasons on cloud. ``` ? github.com/qri-io/qri [no test files]...

The current generation process for the OpenAPI spec in `qri/docs` uses generic placeholders for some of the response structs and should move to generating the full response schemas with reflection.

Since bodies are hard to manage in the suite and generally lead to a lot of fatigue which removes from the precision and usefulness of the tests, we agreed to...

Currently when running `qri registry signup` it asks for a password to complete the process. However it doesn't have a "repeat password" step and can be error prone if you...

There was a discussion how to present an interface for the registry methods which we want to also port over to the core side, but might have differing implementations. We...

Currently we have a fairly basic way to pass in filter & order values. It's either a single "magic" string that's handled internally (as on `lib.ListParams`) or a loosely defined...

Checklist out all api endpoints that need page info but don't have it _Originally posted by @b5 in https://github.com/qri-io/qrimatic/pull/207#discussion_r660078443_

Our auth implementation for a local qri node has very weak guarantees about the actual user engaging with the API. The token provider simply hands out tokens for any correct...

The implementation of the token.Provider interface in qri is stateless and operates purely from the generated tokens. While that is sufficient for day to day usage and operation, in a...