Closed joelanford closed 12 months ago
I like the ideas discussed in that doc. It'll be beneficial to reduce burden for on-cluster components, but I'm curious, is there no better mechanisms today to communicate images other than base64 encoded data for those images? Even when I'm looking at fbc files from a catalog author's perspective, they're clunky. That with the olm.bundle.object
combined together made it extremely hard for me to scroll through fbc metadata and made me think that the file is mostly encoded data (I.e it made it hard for me to focus on the actual metadata).
@joelanford is this still relevant? Now we store this in the filesystem and this information is served over HTTP along with the rest of the catalog contents
It's probably reasonable to close this now that catalogd is just serving the FBC directly.
I do think there's still something to think about here, but in my mind, that something is related to OCI artifacts and I'd imagine that if/when we start thinking about that possibility more, we'll have this aspect covered anyway.
See: https://github.com/operator-framework/operator-lifecycle-manager/blob/master/doc/contributors/design-proposals/operator-logos.md