perf: override default implementation in ArrayIter with dedicated null/non-nullable versions
#8697
Conversation
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
…ull/non-nullable versions implemented for: - `nth` / `nth_back` - `last` - `count` - `for_each` - `fold` / `rfold` - `all` - `any` - `find_map` - `find` / `rfind` - `partition` - `position` / `rposition`
rluvaton
added a commit
to rluvaton/arrow-rs
that referenced
this pull request
Oct 27, 2025
…from the iterator back this also adds a LOT of tests extracted from (which is how I found that bug): - apache#8697
alamb
pushed a commit
that referenced
this pull request
Nov 3, 2025
…from the iterator back (#8728) # Which issue does this PR close? N/A # Rationale for this change for the fix: the array iterator is marked as exact size iterator and double ended iterator so it should report the current length when accessed through the other side # What changes are included in this PR? fix by using `current_end` instead of `array.len()` and also adds a LOT of tests extracted from (which is how I found that bug): - #8697 # Are these changes tested? Yes # Are there any user-facing changes? Kinda
# Conflicts: # arrow-array/src/iterator.rs
Member
Author
|
@alamb this can now be reviewed, it is pretty simple |
Contributor
|
Do we have any performance benchmarks that show if this improves performance? |
Member
Author
|
Did not run any but it should be, I will try to create benchmarks later |
rluvaton
added a commit
to rluvaton/arrow-rs
that referenced
this pull request
Nov 3, 2025
Needed for: - apache#8697
Member
Author
|
@alamb added benchmarks in: |
rluvaton
added a commit
to rluvaton/arrow-rs
that referenced
this pull request
Nov 3, 2025
Needed for: - apache#8697
Member
Author
|
I run the benchmarks and it turned out it is faster in the default implementation for some cases by a lot for some reason, but the |
alamb
pushed a commit
that referenced
this pull request
Nov 11, 2025
… and `count` (#8785) # Which issue does this PR close? N/A # Rationale for this change The default implementations iterate over the iterator to get the value, while we can do that in constant time # What changes are included in this PR? override `nth`, `nth_back`, `last` and `count` # Are these changes tested? existing tests in this file that I added in previous pr # Are there any user-facing changes? Nope ----- Extracted from the following PR as I probably close it as it is not faster locally in some cases: - #8697
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.
Which issue does this PR close?
N/A
Rationale for this change
Improve the performance of array iterator functions by skipping null checks on non nullable
What changes are included in this PR?
override
ArrayIterdefault function implementation with one that check if there are nulls or not:nth/nth_backlastcountfor_eachfold/rfoldallanyfind_mapfind/rfindpartitionposition/rpositionthis implement all functions in
Iterator/DoubleEndedIteratorthat the default implementation is either using some base function that we can't implement in stable (e.g.try_fold) or the implementation is naive (e.g. callingnext()a lot of times)Are these changes tested?
Extracted the tests:
ArrayIterdoes not report size hint correctly after advancing from the iterator back #8728Are there any user-facing changes?
Nope
Benchmarks are in:
ArrayIterbenchmarks #8774