You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "Chen-Yuan-Lai (via GitHub)" <gi...@apache.org> on 2025/02/05 10:19:15 UTC
[PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNu… [datafusion]
Chen-Yuan-Lai opened a new pull request, #14504:
URL: https://github.com/apache/datafusion/pull/14504
…llBufferBuilder
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
Closes #14115 .
## Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
-->
As mentioned in #14115 , several examples in DataFusion codebase still using BooleanBufferBuilder rather than NullBufferBuilder, they should be replaced with NullBufferBuilder for optimization.
## What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
Most methods in `MaybeNullBufferBuilde` can be covered by `NullBufferBuilder`, so I changed the type from `enum` to `struct` which only has one `NullBufferBuilder`.
## Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are they covered by existing tests)?
-->
## Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
-->
<!--
If there are any breaking changes to public APIs, please add the `api change` label.
-->
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #14504:
URL: https://github.com/apache/datafusion/pull/14504#issuecomment-2640789173
I also merged this PR up to main to rerun CI
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #14504:
URL: https://github.com/apache/datafusion/pull/14504#issuecomment-2638241145
Looks like there are some CI failures so marking this PR as a draft
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on code in PR #14504:
URL: https://github.com/apache/datafusion/pull/14504#discussion_r1945294378
##########
datafusion/physical-plan/src/aggregates/group_values/null_builder.rs:
##########
@@ -15,120 +15,78 @@
// specific language governing permissions and limitations
// under the License.
-use arrow::array::BooleanBufferBuilder;
+use arrow::array::NullBufferBuilder;
use arrow::buffer::NullBuffer;
/// Builder for an (optional) null mask
///
/// Optimized for avoid creating the bitmask when all values are non-null
#[derive(Debug)]
-pub(crate) enum MaybeNullBufferBuilder {
- /// seen `row_count` rows but no nulls yet
- NoNulls { row_count: usize },
- /// have at least one null value
- ///
+pub(crate) struct MaybeNullBufferBuilder {
/// Note this is an Arrow *VALIDITY* buffer (so it is false for nulls, true
/// for non-nulls)
- Nulls(BooleanBufferBuilder),
+ nulls: NullBufferBuilder,
}
impl MaybeNullBufferBuilder {
/// Create a new builder
pub fn new() -> Self {
- Self::NoNulls { row_count: 0 }
+ Self {
+ nulls: NullBufferBuilder::new(0),
+ }
}
/// Return true if the row at index `row` is null
pub fn is_null(&self, row: usize) -> bool {
- match self {
- Self::NoNulls { .. } => false,
+ match self.nulls.as_slice() {
// validity mask means a unset bit is NULL
- Self::Nulls(builder) => !builder.get_bit(row),
+ Some(_) => !self.nulls.is_valid(row),
+ None => false,
}
}
/// Set the nullness of the next row to `is_null`
///
- /// num_values is the current length of the rows being tracked
- ///
/// If `value` is true, the row is null.
/// If `value` is false, the row is non null
pub fn append(&mut self, is_null: bool) {
- match self {
- Self::NoNulls { row_count } if is_null => {
- // have seen no nulls so far, this is the first null,
- // need to create the nulls buffer for all currently valid values
- // alloc 2x the need given we push a new but immediately
- let mut nulls = BooleanBufferBuilder::new(*row_count * 2);
- nulls.append_n(*row_count, true);
- nulls.append(false);
- *self = Self::Nulls(nulls);
- }
- Self::NoNulls { row_count } => {
- *row_count += 1;
- }
- Self::Nulls(builder) => builder.append(!is_null),
- }
+ self.nulls.append(!is_null)
}
pub fn append_n(&mut self, n: usize, is_null: bool) {
- match self {
- Self::NoNulls { row_count } if is_null => {
- // have seen no nulls so far, this is the first null,
- // need to create the nulls buffer for all currently valid values
- // alloc 2x the need given we push a new but immediately
- let mut nulls = BooleanBufferBuilder::new(*row_count * 2);
- nulls.append_n(*row_count, true);
- nulls.append_n(n, false);
- *self = Self::Nulls(nulls);
- }
- Self::NoNulls { row_count } => {
- *row_count += n;
- }
- Self::Nulls(builder) => builder.append_n(n, !is_null),
+ if is_null {
+ self.nulls.append_n_nulls(n);
+ } else {
+ self.nulls.append_n_non_nulls(n);
}
}
/// return the number of heap allocated bytes used by this structure to store boolean values
pub fn allocated_size(&self) -> usize {
- match self {
- Self::NoNulls { .. } => 0,
- // BooleanBufferBuilder builder::capacity returns capacity in bits (not bytes)
- Self::Nulls(builder) => builder.capacity() / 8,
- }
+ // NullBufferBuilder builder::allocated_size returns capacity in bits
Review Comment:
Indeed -- I verified this is the case and added some tests here:
- https://github.com/apache/arrow-rs/pull/7089
##########
datafusion/physical-plan/src/aggregates/group_values/null_builder.rs:
##########
@@ -15,120 +15,78 @@
// specific language governing permissions and limitations
// under the License.
-use arrow::array::BooleanBufferBuilder;
+use arrow::array::NullBufferBuilder;
use arrow::buffer::NullBuffer;
/// Builder for an (optional) null mask
///
/// Optimized for avoid creating the bitmask when all values are non-null
#[derive(Debug)]
-pub(crate) enum MaybeNullBufferBuilder {
- /// seen `row_count` rows but no nulls yet
- NoNulls { row_count: usize },
- /// have at least one null value
- ///
+pub(crate) struct MaybeNullBufferBuilder {
/// Note this is an Arrow *VALIDITY* buffer (so it is false for nulls, true
/// for non-nulls)
- Nulls(BooleanBufferBuilder),
+ nulls: NullBufferBuilder,
}
impl MaybeNullBufferBuilder {
/// Create a new builder
pub fn new() -> Self {
- Self::NoNulls { row_count: 0 }
+ Self {
+ nulls: NullBufferBuilder::new(0),
+ }
}
/// Return true if the row at index `row` is null
pub fn is_null(&self, row: usize) -> bool {
- match self {
- Self::NoNulls { .. } => false,
+ match self.nulls.as_slice() {
// validity mask means a unset bit is NULL
- Self::Nulls(builder) => !builder.get_bit(row),
+ Some(_) => !self.nulls.is_valid(row),
+ None => false,
}
}
/// Set the nullness of the next row to `is_null`
///
- /// num_values is the current length of the rows being tracked
- ///
/// If `value` is true, the row is null.
/// If `value` is false, the row is non null
pub fn append(&mut self, is_null: bool) {
- match self {
- Self::NoNulls { row_count } if is_null => {
- // have seen no nulls so far, this is the first null,
- // need to create the nulls buffer for all currently valid values
- // alloc 2x the need given we push a new but immediately
- let mut nulls = BooleanBufferBuilder::new(*row_count * 2);
- nulls.append_n(*row_count, true);
- nulls.append(false);
- *self = Self::Nulls(nulls);
- }
- Self::NoNulls { row_count } => {
- *row_count += 1;
- }
- Self::Nulls(builder) => builder.append(!is_null),
- }
+ self.nulls.append(!is_null)
}
pub fn append_n(&mut self, n: usize, is_null: bool) {
- match self {
- Self::NoNulls { row_count } if is_null => {
- // have seen no nulls so far, this is the first null,
- // need to create the nulls buffer for all currently valid values
- // alloc 2x the need given we push a new but immediately
- let mut nulls = BooleanBufferBuilder::new(*row_count * 2);
- nulls.append_n(*row_count, true);
- nulls.append_n(n, false);
- *self = Self::Nulls(nulls);
- }
- Self::NoNulls { row_count } => {
- *row_count += n;
- }
- Self::Nulls(builder) => builder.append_n(n, !is_null),
+ if is_null {
+ self.nulls.append_n_nulls(n);
+ } else {
+ self.nulls.append_n_non_nulls(n);
}
}
/// return the number of heap allocated bytes used by this structure to store boolean values
pub fn allocated_size(&self) -> usize {
- match self {
- Self::NoNulls { .. } => 0,
- // BooleanBufferBuilder builder::capacity returns capacity in bits (not bytes)
- Self::Nulls(builder) => builder.capacity() / 8,
- }
+ // NullBufferBuilder builder::allocated_size returns capacity in bits
Review Comment:
I actually looked into it more -- I actually think `allocated_size` returns 8* the actual capacity
##########
datafusion/physical-plan/src/aggregates/group_values/null_builder.rs:
##########
@@ -15,120 +15,78 @@
// specific language governing permissions and limitations
// under the License.
-use arrow::array::BooleanBufferBuilder;
+use arrow::array::NullBufferBuilder;
use arrow::buffer::NullBuffer;
/// Builder for an (optional) null mask
///
/// Optimized for avoid creating the bitmask when all values are non-null
#[derive(Debug)]
-pub(crate) enum MaybeNullBufferBuilder {
- /// seen `row_count` rows but no nulls yet
- NoNulls { row_count: usize },
- /// have at least one null value
- ///
+pub(crate) struct MaybeNullBufferBuilder {
/// Note this is an Arrow *VALIDITY* buffer (so it is false for nulls, true
/// for non-nulls)
- Nulls(BooleanBufferBuilder),
+ nulls: NullBufferBuilder,
}
impl MaybeNullBufferBuilder {
/// Create a new builder
pub fn new() -> Self {
- Self::NoNulls { row_count: 0 }
+ Self {
+ nulls: NullBufferBuilder::new(0),
+ }
}
/// Return true if the row at index `row` is null
pub fn is_null(&self, row: usize) -> bool {
- match self {
- Self::NoNulls { .. } => false,
+ match self.nulls.as_slice() {
// validity mask means a unset bit is NULL
- Self::Nulls(builder) => !builder.get_bit(row),
+ Some(_) => !self.nulls.is_valid(row),
+ None => false,
}
}
/// Set the nullness of the next row to `is_null`
///
- /// num_values is the current length of the rows being tracked
- ///
/// If `value` is true, the row is null.
/// If `value` is false, the row is non null
pub fn append(&mut self, is_null: bool) {
- match self {
- Self::NoNulls { row_count } if is_null => {
- // have seen no nulls so far, this is the first null,
- // need to create the nulls buffer for all currently valid values
- // alloc 2x the need given we push a new but immediately
- let mut nulls = BooleanBufferBuilder::new(*row_count * 2);
- nulls.append_n(*row_count, true);
- nulls.append(false);
- *self = Self::Nulls(nulls);
- }
- Self::NoNulls { row_count } => {
- *row_count += 1;
- }
- Self::Nulls(builder) => builder.append(!is_null),
- }
+ self.nulls.append(!is_null)
}
pub fn append_n(&mut self, n: usize, is_null: bool) {
- match self {
- Self::NoNulls { row_count } if is_null => {
- // have seen no nulls so far, this is the first null,
- // need to create the nulls buffer for all currently valid values
- // alloc 2x the need given we push a new but immediately
- let mut nulls = BooleanBufferBuilder::new(*row_count * 2);
- nulls.append_n(*row_count, true);
- nulls.append_n(n, false);
- *self = Self::Nulls(nulls);
- }
- Self::NoNulls { row_count } => {
- *row_count += n;
- }
- Self::Nulls(builder) => builder.append_n(n, !is_null),
+ if is_null {
+ self.nulls.append_n_nulls(n);
+ } else {
+ self.nulls.append_n_non_nulls(n);
}
}
/// return the number of heap allocated bytes used by this structure to store boolean values
pub fn allocated_size(&self) -> usize {
- match self {
- Self::NoNulls { .. } => 0,
- // BooleanBufferBuilder builder::capacity returns capacity in bits (not bytes)
- Self::Nulls(builder) => builder.capacity() / 8,
- }
+ // NullBufferBuilder builder::allocated_size returns capacity in bits
Review Comment:
Nice find
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #14504:
URL: https://github.com/apache/datafusion/pull/14504#issuecomment-2643732450
Pretty stoked about this one
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "Chen-Yuan-Lai (via GitHub)" <gi...@apache.org>.
Chen-Yuan-Lai commented on PR #14504:
URL: https://github.com/apache/datafusion/pull/14504#issuecomment-2639231104
@alamb I found that `b.capacity()` (`BooleanBufferBuilder`) return bits, so `allocated_size()` (`NullBufferBuilder`) should return bits, not bytes
https://docs.rs/arrow-buffer/54.1.0/src/arrow_buffer/builder/null.rs.html#220
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #14504:
URL: https://github.com/apache/datafusion/pull/14504#issuecomment-2643732036
🚀
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] refactor: switch BooleanBufferBuilder to NullBufferBuilder in MaybeNullBufferBuilder [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb merged PR #14504:
URL: https://github.com/apache/datafusion/pull/14504
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org