Feather files

Feather is Arrow’s on-disk format, built for speed rather than long-term storage. AddisDB reads it directly, so a dataframe someone saved from pandas or R is immediately queryable.

Who it is for

Feather exists to move a dataframe between processes and languages without a serialization cost. Writing and reading are close to a memory copy, and the types survive the trip intact — no guessing whether a column came back as a string.

That fidelity is the point. A CSV round-trip loses the difference between an integer and a float, turns dates into text and drops categorical types entirely; Feather preserves all of it because the file is the in-memory layout.

Use it for intermediate results in a data-science pipeline, for passing data between Python and R, and for caching an expensive computation you do not want to recompute.

Producing a file

# pandas
df.to_feather("results.feather")

# R
arrow::write_feather(df, "results.feather")

Feather v2 is the current format and what both of those write by default. A very old v1 file is a different layout — rewrite it from the source rather than trying to read it directly.

Open it in AddisDB

1. New Connection → Files → Feather → Feather.
2. Choose the .feather file, name the connection, and Save.

What AddisDB gives you

• SQL over the dataframe, with the columns and their real types in the sidebar.
• The Chart view for turning an aggregation into a visual.
• Notebooks and AI chat grounded in the file’s columns.
• ⌘K search across every connection, and a file that is only ever read.