Naming related to data files

[SergeiPatiakin]

There are three concepts related to Iceberg data files:

A: files defining rows to add
B: files defining rows to remove
C: files defining either rows to add or rows to remove (encompasses both A and B)

The term "delete file" unambiguously refers to B. On the other hand, "data file" can refer to A or to C. When one sees a variable or field called data_files, it is hard to know whether that refers to A or C. For example at

iceberg-rust/iceberg-rust/src/table/transaction/operation.rs

Line 101 in 2e2cf62

data_files,

it refers to A but at

iceberg-rust/iceberg-rust/src/table/transaction/operation.rs

Line 119 in 2e2cf62

let data_files_iter = delete_files.iter().chain(data_files.iter());

it refers to C.

I wonder if we can come up with distinct canonical terms for A, B, C. E.g.

A: "data file"
B: "delete file"
C: "content file"

Or alternatively:

A: "insert file"
B: "delete file"
C: "data file"

Or any other 3 distinct terms for A, B, C.

Unfortunately the Iceberg spec suffers from the same issue so we cannot use it for inspiration.

[JanKaul]

Good point, I like your first suggestion.