WIP for using page_start_offsets instead of new structure - #2
Open
alamb wants to merge 3 commits into
Open
Conversation
alamb
commented
Jan 27, 2026
alamb
left a comment
alamb
left a comment
Author
There was a problem hiding this comment.
My plan here is to get this PR to compile, then switch to focus on testing, perhaps as a separate PR to main
| let Some(mask_chunk) = | ||
| mask_cursor.next_mask_chunk(batch_size, page_boundaries.as_deref()) | ||
| else { | ||
| let Some(mask_chunk) = mask_cursor.next_mask_chunk(batch_size) else { |
Author
There was a problem hiding this comment.
The idea is that the cursor how has all the information it needs to make the next decision
| /// Optional page start offsets for each requested column. | ||
| /// | ||
| /// See [`FetchRanges::page_start_offsets`] for more details | ||
| page_start_offsets: Option<Vec<Vec<u64>>>, |
Author
There was a problem hiding this comment.
instead of a new struct, I plumbed the offsets all the way through from where the data is requested
Owner
There was a problem hiding this comment.
It seems like these page_start_offsets are byte, not row offsets so we can't really reuse them.
| .with_offset_index_metadata(offset_index_metadata, &self.projection); | ||
|
|
||
| let plan = plan_builder.build(); | ||
| let plan = plan_builder |
Author
There was a problem hiding this comment.
It also simplifies the logic here substantially
sdf-jkl
pushed a commit
that referenced
this pull request
Feb 3, 2026
# Which issue does this PR close? small optimization # Rationale for this change key insight is the byte clone is cheap just a ref count compare to vec clone is a alloc + memcopy. before ``` let mut result = Vec::new(); // alloc #1 result.extend_from_slice(prefix); result.extend_from_slice(suffix); let data = Bytes::from(result.clone()); // alloc #2 + memcpy item.set_from_bytes(data); self.previous_value = result; // keep Vec ``` after ``` let mut result = Vec::with_capacity(prefix_len + suffix.len()); // alloc #1 result.extend_from_slice(&self.previous_value[..prefix_len]); result.extend_from_slice(suffix); let data = Bytes::from(result); // no alloc, takes Vec buffer item.set_from_bytes(data.clone()); // cheap refcount bump self.previous_value = data; // move, no alloc ``` # What changes are included in this PR? previous_value type changed to Bytes preallocate result vec capacity. # Are these changes tested? the existing test should pass # Are there any user-facing changes? no
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This is a WIP targeting @sdf-jkl 's PR
The idea is that we already had the page offsets so we can reuse them when calculating mask offsets
It doesn't yet compile -- I need to update the tests and skipping logic but the basic idea I think is solid