Dan Allan
Dan Allan
P.S. The above imagines a SQLite database `catalog.db` and that is probably best for the vast majority of users. But we could easily support a PostgreSQL target for `download` and...
This will require adding an endpoint like `/assets/{id}`, which is something we wanted anyway.
Good question, this wrinkle had not occurred to me yet. In #450, we use ZIP to bundle multiple buffers (numpy arrays) into one response. We considered TAR, but ZIP has...
I wonder if `description` sounds too much like it would be the return value of the existing method `describe()`.
To finish putting my thoughts on paper here: Making these serializer functions play two different roles (return/yield bytes vs write bytes into file) is not my favorite thing. An alternative...
Elaborating on the above: ```py def serialize_csv_sync(df, metadata): ... yield chunk_of_bytes async def serialize_csv_async(df, metadata, file): ... file.write(...) # return None serialization_registry.register("array", "text/csv", serialize_csv_sync, serialize_csv_async) ```
Another point we should consider in this API design: * What about formats like Zarr and TileDB that span multiple files? This is another reason, along with performance optimization for...
There is renewed interest in this, prompted by @EliotGann.
Thinking on this for awhile, I lean toward: * Always enforce (1) * Provide a configurable boolean switch to opt in to (2).
In #327 we enforced that each `spec` has to be declared, but having a `validator` is optional. Still not sure if we should do anything with (2).