We need help - abstractfactory/openmetadata GitHub Wiki

This page summarises some of the things we're looking for help with.

Binary format

The Open Metadata data-store currently consists of files and folders, but for performance and atomicity reasons we are also looking into the prospect of storing data into a binary, hierarchical format; such as Alembic or HDF5.

Concurrency

Data written is currently on a "first come, first served" basis, meaning there exists the potential of multiple writes onto a single location at once. For individual users, this isn't an issue, but for multi-user situations, such as when involved in feature film-level content creation of hundreds or more users this could potentially cause an amount of headaches.

One of the considerations we've made is to centralise calls writing into the data-store so as to facilitate fair-queuing and a single thread of communication. This was, no two writes can happen. The distribution could made through a RabbitMQ-style broker or similar.

Versioning

There currently exists no versioning scheme for metadata stored with Open Metadata. This may cause trouble in cases of upgrading to newer versions, or when attempting to access an older data-store with a newer library.

General Development

We're always on the lookout for those interested in decentralised metadata for content creation. If that's you, drop us a line, [email protected]