Reading through the Zero docs on authentication, I saw that JWT authentication was all that was supported (for now). I'm building a project using Ruby on Rails as an API and Next.js for the UI. While JWT auth is possible with Rails, most auth implementations are session based, included the built in auth solution to the latest Rails 8 release. Thus I wanted to figure out how to best set up Zero while still using the built in session based auth.

In this reply @aa, outlined how the Zero constructor can take a function for the auth param https://bsky.app/profile/aaronboodman.com/post/3ldyz5bet3s2s

This all sounds right. The auth param to Zero's constructor can be a function. When the token expires, Zero invokes the function to get a new token. It's async so you can call an endpoint or whatever.

and later

To clarify further:

• keep the refresh token in an http-only cookie

Collecting thoughts and information about background jobs across many different programming languages. What are commonalities, differences, etc. The list definitely won't be exhaustive, it just needs enough information to make some decisions.

• Sidekiq - https://github.com/sidekiq/sidekiq
• SolidQueue - https://github.com/rails/solid_queue
• Job Iteration - https://github.com/Shopify/job-iteration
• Built ontop of previous background job implementations to add durability to jobs

I wanted to figure out how I might be able to take advantage of Apollo Client caching while using RSC. I came up with this approach, which is rather rudimentary, but seems like a good starting point. It is taking advantage of the new localStorage API added in Node 22 https://nodejs.org/api/globals.html#localstorage

I built up a new query method which wraps the existing query and adds in ability to pass in an identifier (likely a user ID or something of that matter) to ensure that caches don't overlap.