HadrienG2 opened a new pull request, #5754: URL: https://github.com/apache/arrow-rs/pull/5754
# Which issue does this PR close? Closes #5749. # Rationale for this change This makes the interface of different builder types more consistent, and enables exposing builder contents as a slice in more cases in my prototype for #5700. # What changes are included in this PR? Overall, my general process was to look at what `append_null()`/`append(false)` pushes the null into, and expose the corresponding null buffer. - For each builder which has an inner null buffer builder, it is exposed as already done in `BooleanBuilder` and `PrimitiveBuilder`. - For builders that use a dictionary-like (key, value) layout, the null buffer of keys is treated as the overall null buffer of the builder, following the example of `append_null()`. This only leaves the following builders **without** a null buffer accessor after this PR: - `XyzRunBuilder`: For these builder types, the null buffer is per-run, not per-element. It is not yet clear to me if/how that should be exposed, so I'm leaving this for a future PR. - `UnionBuilder`: For this builder type, each variant has its own null buffer, so there is no notion of builder-wide null buffer. This would be best handled by exposing some sort of access to the storage of individual variants, but again, this requires more design work, so I'm leaving it for a future PR. # Are there any user-facing changes? More builders expose a `validity_slice(&self) -> Option<&[u8]>` method. Since the semantics are identical to pre-existing methods from other builders with this name, the documentation was copy-pasted from there. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
