DataFusion weekly project plan (Andrew Lamb) - Sep 9, 2024

[alamb]

Follow on to #12336

My (personal) North ⭐ : 1000 projects are built using DataFusion 📈 ( Getting closer all the time )

It would be great for other contributors to DataFusion to to make them visible somehow as well 🙏 (feel free to copy / modify the format of this ticket)

External news

Major Discussions (Please Take a Look)

• What features should DataFusion have? [DISCUSS] Document criteria for adding new features / what belongs in core DataFusion (e.g. sql syntax, functions, etc) #12357

Upcoming Meetups

• San Francisco datafusion meetup on Wednesday September 25th https://lu.ma/05lyf19h
• New York, NY, USA DataFusion Meetup: Tuesday Sep 17th 2024, 5:30PM - 8:30PM Manhattan #11213
• DataFusion Belgrade Meetup 2024/09/27 #11431

Highlights from last week

• @eejbyfeldt improved performance of filtering with feat: Add projection to FilterExec #12281
• @2010YOUY01 began optimizing string functions by special casing ASCII Faster character_length() string function for ASCII-only case #12356
• @haohuaijin fixed a regression in planning performance: perf: avoid repeat format in calc_func_dependencies_for_project #12305
• @sgrebnov @phillipleblanc and others have been making the SQL unparser better: Fix subquery alias table definition unparsing for SQLite #12331, etc
• @Dandandan is making joins even better: Push down null filters for more join types #12348
• Thanks to @devanbenz @wiedld , @Kev1n8 and others we are getting close to ready for StringView (e.g. feat: Add projection to FilterExec #12281, Update the CONCAT scalar function to support Utf8View #12224, Improve StringView support for SUBSTR #12044)
• Updated to the latest arrow/parquet version: Update to arrow/parquet 53.0.0, tonic, prost, object_store, pyo3 #12032
• @findepi removed Expr::Sort (this was 2 weeks ago technically, but I didn't write one of these lists last week): Remove Sort expression (Expr::Sort) #12177

My (personal) plans for this week

• complete board report September 2024 ASF Board Report #10156
• Work down the review backlog

Projects I plan to help actively help review / push forward

• StringView integration [Epic] Complete Initial StringView in DataFusion #11752
• Parquet Filtering: Return TableProviderFilterPushDown::Exact when Parquet Pushdown Enabled #4028
• High Cardinality Grouping: [Epic] High cardinality aggregation performance wishlist #11679

Algorithm for (my) prioritizing PR reviews

Note there are many committers who can and do review and merge PRs, so this is not the priorities of the project as a whole, just the approximate algorithm I use to prioritize my own time.

Priority:

1. Bug fixes (where something is just incorrect), especially regressions (where it used to work and now does not)
2. Improvements directly related to features needed for InfluxDB (my employer)
3. Documentation and test improvements (I view these as very strategically important)
4. PRs for other features I think are strategically important (See below)
5. Other new features / additions to functionality (note this is the lowest priority on purpose)

The top strategically important projects in my head are:

• Anything that improves DataFusion's quality such as bug fixes, and improved / expanded tests, etc
• Anything that makes it easier to use DataFusion as a user (docs, examples, better APIs, cleaner code, etc), including Getting started guide for new users (who want to use DataFusion in their project)
• Improved performance (with benchmarks)

Thus, if you are interested in contributing to DataFusion and are interested in a fast turn around time I would recommend looking into bug fixes / test improvements / documentation or the projects named above.

If you propose adding new functionality, especially if the PR is large/complex and not connected to a wider need, the review cycle will likely be longer. You can make it a shorter cycle by looking at the comments on other recent PRs and following the same model (e.g. break it into small PRs, ensure there are tests in sqllogictest for example, the CI passes, includes documentation, etc)

Background

The idea of this ticket is make my plans for DataFusion visible, largely for my own personal organizational needs, but also to:

1. Communicate / coordinate in the community
2. Help provide an interesting summary of what is happening in DataFusion this week (both for my future self and others)

[alamb]

Review Queue

DataFusion (bugs / improvements)

• Fix incorrect OFFSET during LIMIT pushdown. #12399
• Support timestamps and steps of less than a day for range/generate_series #12400

DataFusion (performance):

• Enable datafusion.optimizer.filter_null_join_keys by default #12369
• Avoid RowConverter for multi column grouping (10% faster clickbench queries) #12269
• Impl convert_to_state for GroupsAccumulatorAdapter (faster median for high cardinality aggregates) #11827
• Improve performance of trim for string view (10%) #12395
• Faster strpos() string function for ASCII-only case #12401
• Sketch for aggregation intermediate results blocked management #11943

DataFusion (new features, lower priority for me):

• Adding node_id to ExecutionPlanProperties #12186
• Implement SHOW FUNCTIONS #12266
• Faster strpos() string function for ASCII-only case #12401
• include input fields as output for Substrait consumer #12225
• Add additional regexp function regexp_count() #12080
• Support skewness(x) in Aggregation function #12295
• Remove LargeUtf8|Binary, Utf8|BinaryView, and Dictionary from ScalarValue #11978

Arrow-rs

• Support StringViewArray interop with python: fix lingering C Data Interface issues for *ViewArray arrow-rs#6368
• Implement native support StringViewArray for regexp_is_match and regexp_is_match_scalar function, deprecate regexp_is_match_utf8 and regexp_is_match_utf8_scalar arrow-rs#6376
• Improve performance of set_bits by avoiding to set individual bits arrow-rs#6288
• Make unary and binary faster arrow-rs#6365
• Parquet: Verify 32-bit CRC checksum when decoding pages arrow-rs#6290
• feat: add Allocator type param to MutableBuffer arrow-rs#6336

sqlparser-rs

• With Order Support for Memory Tables datafusion-sqlparser-rs#1401

[emgeee]

Also have an upcoming SF datafusion meetup on Wednesday September 25th https://lu.ma/05lyf19h

[alamb]

Also have an upcoming SF datafusion meetup on Wednesday September 25th https://lu.ma/05lyf19h

Thanks, sorry added it to the description

[alamb]

Next week: #12494