GH-39803: [C++][Acero] Fix AsOfJoin with differently ordered schemas than the output #39804
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
|
|
pitrou
changed the title
pitrou
reviewed
Feb 12, 2024
| @@ -16,6 +16,7 @@ | |||
| // under the License. | |||
|
|
|||
| #include <gmock/gmock-matchers.h> | |||
| #include <iostream> // nocommit | |||
There was a problem hiding this comment.
Is this required? Also, what is "nocommit" for?
There was a problem hiding this comment.
whoops - this was for some cout debugging, will remove
|
Also, can you please rebase from git main? |
github-actions
bot
added
awaiting committer review
JerAguilon
force-pushed
the
fix-schema
branch
from
eaffb69
to
628dd0a
Compare
Done and addressed comments. Thanks for the review |
pitrou
reviewed
Feb 19, 2024
pitrou
approved these changes
Feb 19, 2024
|
@github-actions crossbow submit -g cpp |
|
Revision: 11436fd Submitted crossbow builds: ursacomputing/crossbow @ actions-4147b43cbc
|
|
Thanks a lot for this @JerAguilon ! |
|
After merging your PR, Conbench analyzed the 6 benchmarking runs that have been run so far on merge-commit 4b74b45. There were no benchmark performance regressions. 🎉 The full Conbench report has more details. It also includes information about 9 possible false positives for unstable benchmarks that are known to sometimes produce them. |
raulcd
pushed a commit
that referenced
this pull request
Feb 20, 2024
…than the output (#39804) ### Rationale for this change Issue is described visually in #39803. The key hasher works by hashing every row of the input tables' key columns. An important step is inspecting the [column metadata](https://github.com/apache/arrow/blob/main/cpp/src/arrow/acero/asof_join_node.cc#L412) for the asof-join key fields. This returns whether columns are fixed width, among other things. The issue is we are passing the `output_schema`, rather than the input's schema. If an input looks like ``` key_string_type,ts_int32_type,val ``` But our expected output schema looks like: ``` ts_int32,key_string_type,... ``` Then the hasher will think that the `key_string_type`'s type is an int32. This completely throws off hashes. Tests currently get away with it since we just use ints across the board. ### What changes are included in this PR? One line fix and test with string types. ### Are these changes tested? Yes. Can see the test run before and after changes here: https://gist.github.com/JerAguilon/953d82ed288d58f9ce24d1a925def2cc Before the change, notice that inputs 0 and 1 have mismatched hashes: ``` AsofjoinNode(0x16cf9e2d8): key hasher 1 got hashes [0, 9784892099856512926, 1050982531982388796, 10763536662319179482, 2029627098739957112, 11814237723602982167, 3080328155728858293, 12792882290360550483, 4058972722486426609, 13771526852823217039] ... AsofjoinNode(0x16cf9dd18): key hasher 0 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ``` And after, they do match: ``` AsofjoinNode(0x16f2ea2d8): key hasher 1 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ... AsofjoinNode(0x16f2e9d18): key hasher 0 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ``` ...which is exactly what you want, since the `key` column for both tables looks like `["0", "1", ..."9"]` ### Are there any user-facing changes? * Closes: #39803 Lead-authored-by: Jeremy Aguilon <jeraguilon@gmail.com> Co-authored-by: Antoine Pitrou <pitrou@free.fr> Signed-off-by: Antoine Pitrou <antoine@python.org>
zanmato1984
pushed a commit
to zanmato1984/arrow
that referenced
this pull request
Feb 28, 2024
…hemas than the output (apache#39804) ### Rationale for this change Issue is described visually in apache#39803. The key hasher works by hashing every row of the input tables' key columns. An important step is inspecting the [column metadata](https://github.com/apache/arrow/blob/main/cpp/src/arrow/acero/asof_join_node.cc#L412) for the asof-join key fields. This returns whether columns are fixed width, among other things. The issue is we are passing the `output_schema`, rather than the input's schema. If an input looks like ``` key_string_type,ts_int32_type,val ``` But our expected output schema looks like: ``` ts_int32,key_string_type,... ``` Then the hasher will think that the `key_string_type`'s type is an int32. This completely throws off hashes. Tests currently get away with it since we just use ints across the board. ### What changes are included in this PR? One line fix and test with string types. ### Are these changes tested? Yes. Can see the test run before and after changes here: https://gist.github.com/JerAguilon/953d82ed288d58f9ce24d1a925def2cc Before the change, notice that inputs 0 and 1 have mismatched hashes: ``` AsofjoinNode(0x16cf9e2d8): key hasher 1 got hashes [0, 9784892099856512926, 1050982531982388796, 10763536662319179482, 2029627098739957112, 11814237723602982167, 3080328155728858293, 12792882290360550483, 4058972722486426609, 13771526852823217039] ... AsofjoinNode(0x16cf9dd18): key hasher 0 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ``` And after, they do match: ``` AsofjoinNode(0x16f2ea2d8): key hasher 1 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ... AsofjoinNode(0x16f2e9d18): key hasher 0 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ``` ...which is exactly what you want, since the `key` column for both tables looks like `["0", "1", ..."9"]` ### Are there any user-facing changes? * Closes: apache#39803 Lead-authored-by: Jeremy Aguilon <jeraguilon@gmail.com> Co-authored-by: Antoine Pitrou <pitrou@free.fr> Signed-off-by: Antoine Pitrou <antoine@python.org>
thisisnic
pushed a commit
to thisisnic/arrow
that referenced
this pull request
Mar 8, 2024
…hemas than the output (apache#39804) ### Rationale for this change Issue is described visually in apache#39803. The key hasher works by hashing every row of the input tables' key columns. An important step is inspecting the [column metadata](https://github.com/apache/arrow/blob/main/cpp/src/arrow/acero/asof_join_node.cc#L412) for the asof-join key fields. This returns whether columns are fixed width, among other things. The issue is we are passing the `output_schema`, rather than the input's schema. If an input looks like ``` key_string_type,ts_int32_type,val ``` But our expected output schema looks like: ``` ts_int32,key_string_type,... ``` Then the hasher will think that the `key_string_type`'s type is an int32. This completely throws off hashes. Tests currently get away with it since we just use ints across the board. ### What changes are included in this PR? One line fix and test with string types. ### Are these changes tested? Yes. Can see the test run before and after changes here: https://gist.github.com/JerAguilon/953d82ed288d58f9ce24d1a925def2cc Before the change, notice that inputs 0 and 1 have mismatched hashes: ``` AsofjoinNode(0x16cf9e2d8): key hasher 1 got hashes [0, 9784892099856512926, 1050982531982388796, 10763536662319179482, 2029627098739957112, 11814237723602982167, 3080328155728858293, 12792882290360550483, 4058972722486426609, 13771526852823217039] ... AsofjoinNode(0x16cf9dd18): key hasher 0 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ``` And after, they do match: ``` AsofjoinNode(0x16f2ea2d8): key hasher 1 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ... AsofjoinNode(0x16f2e9d18): key hasher 0 got hashes [17528465654998409509, 12047706865972860560, 18017664240540048750, 12358837084497432044, 8151160321586084686, 8691136767698756332, 15973065724125580046, 9654919479117127288, 618127929167745505, 3403805303373270709] ``` ...which is exactly what you want, since the `key` column for both tables looks like `["0", "1", ..."9"]` ### Are there any user-facing changes? * Closes: apache#39803 Lead-authored-by: Jeremy Aguilon <jeraguilon@gmail.com> Co-authored-by: Antoine Pitrou <pitrou@free.fr> Signed-off-by: Antoine Pitrou <antoine@python.org>
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.
Rationale for this change
Issue is described visually in #39803.
The key hasher works by hashing every row of the input tables' key columns. An important step is inspecting the column metadata for the asof-join key fields. This returns whether columns are fixed width, among other things.
The issue is we are passing the
output_schema, rather than the input's schema.If an input looks like
But our expected output schema looks like:
Then the hasher will think that the
key_string_type's type is an int32. This completely throws off hashes. Tests currently get away with it since we just use ints across the board.What changes are included in this PR?
One line fix and test with string types.
Are these changes tested?
Yes. Can see the test run before and after changes here: https://gist.github.com/JerAguilon/953d82ed288d58f9ce24d1a925def2cc
Before the change, notice that inputs 0 and 1 have mismatched hashes:
And after, they do match:
...which is exactly what you want, since the
keycolumn for both tables looks like["0", "1", ..."9"]Are there any user-facing changes?