woile.eu woile.eu Hi there,
I'm trying to get the .sequoia-state.json in the CI in order to make sequoia inject work.
But both publish and sync seem to require auth, which I was hoping I could avoid.
Right now I run the publish locally, and I want the CI to just inject.
I was wondering if sync really needs the auth, or if it could have a --just-fetch or --fetch-only.
Thanks for the work, I'm getting familiar with it and it looks awesome
sequoia init puts the .sequoia-state.json in the .gitignore, not being an expert on AT, I leave it there. It leads me to belive it shouldn't be tracked, hence me wondering if it's actually needed.https://tangled.org/stevedylan.dev/sequoia/blob/main/packages/cli/src/commands/init.ts#L379-380
Auth is still required unless you provide your PDS URL. Is that something you'd be willing to do, say, as an env var during publishing?
Sync pulls records from your PDS based on your account, so I guess it wouldn't technically need to authenticate if you provided your PDS URI; but, what would be the point in having the CI sync
.sequoia-state.json? I'm trying to understand the workflow here and whether changes are needed. Generally a CI wouldn't make repo changes anyway. It's rare to see that. Perhaps another workflow if you want to automate it is to add a secure variable (app password you can create with your PDS e.g., Bluesky) to your CI and have it publish records on your behalf and then throw away (ignore) it's changes to the state file. You can runsequoia synclocally to update it, but it's not strictly needed. Just takes longer to discover the true state of the collection on your PDS.And if you'd have your CI publish on your behalf with an authenticated token, you could still run
sequoia synctoday.