Fixed Dremel to properly shred/assemble nested structures with nullable elements #778
Conversation
This file contains 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
norberttech
commented
Nov 11, 2023
| @@ -103,25 +121,4 @@ public function test_dremel_shred_on_repeated_columns() : void | |||
| ] | |||
| ); | |||
| } | |||
|
|
|||
| public function test_dremel_with_rows_with_combined_scalar_values_and_array_values() : void | |||
There was a problem hiding this comment.
This constraint was incorrect since there is a scenario where arrays also contain nulls, not only scalar values.
Flow PHP - BenchmarksResults of the benchmarks from this PR are compared with the results from 1.x branch. Extractors+-----------------------+-------------------+------+-----+------------------+-------------------+-----------------+
| benchmark | subject | revs | its | mem_peak | mode | rstdev |
+-----------------------+-------------------+------+-----+------------------+-------------------+-----------------+
| AvroExtractorBench | bench_extract_10k | 1 | 3 | 35.271mb +0.00% | 538.754ms +46.99% | ±2.45% +189.05% |
| CSVExtractorBench | bench_extract_10k | 1 | 3 | 4.837mb +0.01% | 438.088ms +62.60% | ±0.99% +409.14% |
| JsonExtractorBench | bench_extract_10k | 1 | 3 | 5.007mb +0.01% | 857.833ms +44.48% | ±3.21% +172.24% |
| ParquetExtractorBench | bench_extract_10k | 1 | 3 | 233.597mb +2.61% | 1.209s +54.30% | ±1.61% +288.61% |
| TextExtractorBench | bench_extract_10k | 1 | 3 | 4.831mb +0.01% | 25.594ms +32.32% | ±1.04% -28.38% |
| XmlExtractorBench | bench_extract_10k | 1 | 3 | 4.831mb +0.01% | 639.734ms +58.02% | ±0.46% -62.15% |
+-----------------------+-------------------+------+-----+------------------+-------------------+-----------------+
Transformers+-----------------------------+--------------------------+------+-----+-----------------+------------------+---------------+
| benchmark | subject | revs | its | mem_peak | mode | rstdev |
+-----------------------------+--------------------------+------+-----+-----------------+------------------+---------------+
| RenameEntryTransformerBench | bench_transform_10k_rows | 1 | 3 | 87.098mb +0.00% | 82.279ms +66.81% | ±1.14% -6.46% |
+-----------------------------+--------------------------+------+-----+-----------------+------------------+---------------+
Loaders+--------------------+----------------+------+-----+------------------+-------------------+-----------------+
| benchmark | subject | revs | its | mem_peak | mode | rstdev |
+--------------------+----------------+------+-----+------------------+-------------------+-----------------+
| AvroLoaderBench | bench_load_10k | 1 | 3 | 95.281mb +0.00% | 857.591ms +48.70% | ±2.04% +104.61% |
| CSVLoaderBench | bench_load_10k | 1 | 3 | 46.114mb +0.00% | 82.650ms +17.13% | ±0.61% -65.91% |
| JsonLoaderBench | bench_load_10k | 1 | 3 | 90.558mb +0.00% | 93.111ms +46.61% | ±1.19% +9.18% |
| ParquetLoaderBench | bench_load_10k | 1 | 3 | 286.030mb +2.23% | 2.463s +87.11% | ±1.21% +52.68% |
| TextLoaderBench | bench_load_10k | 1 | 3 | 16.608mb +0.00% | 43.523ms +7.33% | ±1.22% +141.73% |
+--------------------+----------------+------+-----+------------------+-------------------+-----------------+
Building Blocks+-------------------------+----------------------------+------+-----+-----------------+-------------------+-------------------------------+
| benchmark | subject | revs | its | mem_peak | mode | rstdev |
+-------------------------+----------------------------+------+-----+-----------------+-------------------+-------------------------------+
| RowsBench | bench_chunk_10_on_10k | 2 | 3 | 60.707mb +0.00% | 5.616ms +151.11% | ±1.25% -39.90% |
| RowsBench | bench_diff_left_1k_on_10k | 2 | 3 | 80.499mb +0.00% | 225.495ms +48.52% | ±0.62% -71.82% |
| RowsBench | bench_diff_right_1k_on_10k | 2 | 3 | 59.025mb +0.00% | 20.509ms +35.79% | ±1.28% +139.97% |
| RowsBench | bench_drop_1k_on_10k | 2 | 3 | 59.846mb +0.00% | 3.071ms +74.09% | ±2.95% +163.78% |
| RowsBench | bench_drop_right_1k_on_10k | 2 | 3 | 59.846mb +0.00% | 3.197ms +66.44% | ±3.17% +92.95% |
| RowsBench | bench_entries_on_10k | 2 | 3 | 59.059mb +0.00% | 4.767ms +73.34% | ±1.95% +47.36% |
| RowsBench | bench_filter_on_10k | 2 | 3 | 59.588mb +0.00% | 24.109ms +71.44% | ±3.39% +173.25% |
| RowsBench | bench_find_on_10k | 2 | 3 | 59.588mb +0.00% | 23.871ms +68.72% | ±2.92% +604.03% |
| RowsBench | bench_find_one_on_10k | 10 | 3 | 57.659mb +0.00% | 2.206μs +16.11% | ±2.11% +18061489817610000.00% |
| RowsBench | bench_first_on_10k | 10 | 3 | 57.659mb +0.00% | 0.400μs 0.00% | ±0.00% 0.00% |
| RowsBench | bench_flat_map_on_1k | 2 | 3 | 65.892mb +0.00% | 16.885ms +61.12% | ±3.42% +67.57% |
| RowsBench | bench_map_on_10k | 2 | 3 | 91.412mb +0.00% | 73.053ms +54.02% | ±2.54% +506.37% |
| RowsBench | bench_merge_1k_on_10k | 2 | 3 | 60.109mb +0.00% | 3.261ms +66.17% | ±1.09% -26.66% |
| RowsBench | bench_partition_by_on_10k | 2 | 3 | 62.379mb +0.00% | 59.537ms +77.61% | ±2.45% +3531.50% |
| RowsBench | bench_remove_on_10k | 2 | 3 | 62.209mb +0.00% | 9.349ms +96.04% | ±2.84% +83.45% |
| RowsBench | bench_sort_asc_on_1k | 2 | 3 | 57.659mb +0.00% | 61.812ms +63.11% | ±0.82% +16.28% |
| RowsBench | bench_sort_by_on_1k | 2 | 3 | 57.659mb +0.00% | 64.372ms +70.50% | ±0.93% -21.45% |
| RowsBench | bench_sort_desc_on_1k | 2 | 3 | 57.659mb +0.00% | 63.206ms +65.30% | ±1.91% -28.01% |
| RowsBench | bench_sort_entries_on_1k | 2 | 3 | 59.933mb +0.00% | 12.215ms +67.36% | ±3.52% +369.21% |
| RowsBench | bench_sort_on_1k | 2 | 3 | 57.658mb +0.00% | 42.951ms +50.60% | ±2.90% +99.45% |
| RowsBench | bench_take_1k_on_10k | 10 | 3 | 57.659mb +0.00% | 25.596μs +97.20% | ±2.41% +85.34% |
| RowsBench | bench_take_right_1k_on_10k | 10 | 3 | 57.659mb +0.00% | 37.688μs +144.63% | ±0.25% -17.98% |
| RowsBench | bench_unique_on_1k | 2 | 3 | 80.500mb +0.00% | 223.549ms +42.78% | ±1.20% +307.85% |
| NativeEntryFactoryBench | bench_entry_factory | 1 | 3 | 93.715mb +0.00% | 203.846ms +64.64% | ±1.07% +166.39% |
| NativeEntryFactoryBench | bench_entry_factory | 1 | 3 | 48.632mb +0.00% | 109.705ms +75.31% | ±3.51% +98.71% |
| NativeEntryFactoryBench | bench_entry_factory | 1 | 3 | 12.653mb +0.01% | 24.297ms +62.36% | ±0.88% -67.86% |
+-------------------------+----------------------------+------+-----+-----------------+-------------------+-------------------------------+
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Change Log
Added
Fixed
Changed
Removed
Deprecated
Security
Description
While working on Dremel algorithms implementation I was mostly focused on non nullable nested structures, that's how this bug was missed in the first implementation.
After this fix, it will not only properly handle maps/lists with nulls but also empty maps/lists.
Dremel is no longer working as a generator because all data needs to sit in the memory anyway and since we are operating at small integers, this was bringing very little value compared to complexity of debugging it.