ARROW-15382: [Java] fix SplitAndTransfer throws for (0,0) if vector empty #12250

wzx140 · 2022-01-25T06:37:07Z

No description provided.

github-actions · 2022-01-25T06:39:15Z

https://issues.apache.org/jira/browse/ARROW-15382

…mpty

wzx140 · 2022-05-18T07:07:13Z

@lidavidm Could you please review it? Really thanks!

lidavidm · 2022-05-18T12:33:37Z

@davisusanibar would you like to take a look here first?

lidavidm

Thanks. I think it would be good to have a test with a non-empty vector but transferring 0 length as well (also, trying a non-zero startIndex but with 0 length).

pitrou · 2022-06-07T11:49:30Z

@wzx140 Would you like to add another unit test as suggested by @lidavidm ?

lidavidm · 2022-11-07T15:03:43Z

@davisusanibar or @lwhite1 would you like to perhaps carry this over the finish line?

lwhite1 · 2022-11-07T19:01:59Z

@lidavidm yes, I will add the suggested tests.

lwhite1 · 2022-11-16T20:47:46Z

Thanks. I think it would be good to have a test with (1) a non-empty vector but transferring 0 length as well (also, (2) trying a non-zero startIndex but with 0 length).

I created a test with a non-empty vector and transferring length 0, but it throws an exception when the allocator is closed java.lang.IllegalStateException: Allocator[ROOT] closed with outstanding buffers allocated (1).

The same test transferring length 1 passes.

lwhite1

As mentioned in earlier comment, this PR does not address situations where the source vector is non-empty, and the transfer length is 0.

amol- · 2023-03-30T17:17:58Z

Closing because it has been untouched for a while, in case it's still relevant feel free to reopen and move it forward 👍

FiV0 · 2024-04-02T09:44:08Z

Hey, could this be reopened. This is still an issue in Arrow 15.0.2 and I run into it multiple times. Also happy to look into it if further work needs to be done.

#41066) This is addresses https://issues.apache.org/jira/browse/ARROW-15382 and is reopening of #12250 (which I asked to be reopened). I tried to address all the comments from the previous discussion, added some more tests and fixed an issue in the old commit. * GitHub Issue: #30866 Authored-by: Finn Völkel <finn.volkel@gmail.com> Signed-off-by: David Li <li.davidm96@gmail.com>

…r empty (apache#41066) This is addresses https://issues.apache.org/jira/browse/ARROW-15382 and is reopening of apache#12250 (which I asked to be reopened). I tried to address all the comments from the previous discussion, added some more tests and fixed an issue in the old commit. * GitHub Issue: apache#30866 Authored-by: Finn Völkel <finn.volkel@gmail.com> Signed-off-by: David Li <li.davidm96@gmail.com>

…ixes. (#81) * apacheGH-30866: [Java] fix SplitAndTransfer throws for (0,0) if vector empty (apache#41066) This is addresses https://issues.apache.org/jira/browse/ARROW-15382 and is reopening of apache#12250 (which I asked to be reopened). I tried to address all the comments from the previous discussion, added some more tests and fixed an issue in the old commit. * GitHub Issue: apache#30866 Authored-by: Finn Völkel <finn.volkel@gmail.com> Signed-off-by: David Li <li.davidm96@gmail.com> * apacheGH-43463: [C++][Gandiva] Always use gdv_function_stubs.h in context_helper.cc (apache#43464) ### Rationale for this change `gdv_function_stubs.h` has declarations of functions in `context_helper.cc`. If we don't include `gdv_function_stubs.h`, it causes attribution mismatch error with unity build. ### What changes are included in this PR? Always include `gdv_function_stubs.h` in `context_helper.cc`. ### Are these changes tested? Yes. ### Are there any user-facing changes? No. * GitHub Issue: apache#43463 Authored-by: Sutou Kouhei <kou@clear-code.com> Signed-off-by: Sutou Kouhei <kou@clear-code.com> * apacheGH-43119: [CI][Packaging] Update manylinux 2014 CentOS repos that have been deprecated (apache#43121) ### Rationale for this change Jobs are failing to find mirrorlist.centos.org ### What changes are included in this PR? Updating repos based on solution from: apache#43119 (comment) ### Are these changes tested? Via archery ### Are there any user-facing changes? No * GitHub Issue: apache#43119 Lead-authored-by: Raúl Cumplido <raulcumplido@gmail.com> Co-authored-by: Sutou Kouhei <kou@clear-code.com> Co-authored-by: Sutou Kouhei <kou@cozmixng.org> Signed-off-by: Raúl Cumplido <raulcumplido@gmail.com> * Update macos deployment target to 12 to match build machine. * apacheGH-43400: [C++] Ensure using bundled GoogleTest when we use bundled GoogleTest (apache#43465) ### Rationale for this change If we use bundled GoogleTest and system other dependencies such as Boost, our include path options may be: * `-isystem /opt/homebrew/include` (for Boost) * `-isystem build_dir/_deps/googletest-src/googletest` (for bundled GoogleTest) * `-isystem build_dir/_deps/googletest-src/googlemock` (for bundled GoogleTest) With this order, GoogleTest headers in `/opt/homebrew/include/` are used with bundled GoogleTest. It may cause link errors. ### What changes are included in this PR? This change introduces a new CMake target `arrow::GTest::gtest_headers` that has include paths for bundled GoogleTest. And it's always used as the first link library of all test program. With this change, our include path options are: * `-isystem build_dir/_deps/googletest-src/googletest` (for bundled GoogleTest) * `-isystem build_dir/_deps/googletest-src/googlemock` (for bundled GoogleTest) * `-isystem /opt/homebrew/include` (for Boost) With this order, we can always use our bundled GoogleTest. `arrow::GTest::gtest_headers` is defined only when we use bundled GoogleTest. So this doesn't change the system GoogleTest case. ### Are these changes tested? Yes. ### Are there any user-facing changes? Yes. * GitHub Issue: apache#43400 Authored-by: Sutou Kouhei <kou@clear-code.com> Signed-off-by: Jacob Wujciak-Jens <jacob@wujciak.de> --------- Signed-off-by: David Li <li.davidm96@gmail.com> Signed-off-by: Sutou Kouhei <kou@clear-code.com> Signed-off-by: Raúl Cumplido <raulcumplido@gmail.com> Signed-off-by: Jacob Wujciak-Jens <jacob@wujciak.de> Co-authored-by: Finn Völkel <FiV0@users.noreply.github.com> Co-authored-by: Sutou Kouhei <kou@clear-code.com> Co-authored-by: Raúl Cumplido <raulcumplido@gmail.com> Co-authored-by: Sutou Kouhei <kou@cozmixng.org>

…r empty (apache#41066) This is addresses https://issues.apache.org/jira/browse/ARROW-15382 and is reopening of apache#12250 (which I asked to be reopened). I tried to address all the comments from the previous discussion, added some more tests and fixed an issue in the old commit. * GitHub Issue: apache#30866 Authored-by: Finn Völkel <finn.volkel@gmail.com> Signed-off-by: David Li <li.davidm96@gmail.com>

github-actions bot added the Component: Java label Jan 25, 2022

ARROW-15382: [Java] fix SplitAndTransfer throws for (0,0) if vector e…

e0fd370

…mpty

wzx140 force-pushed the ARROW-15382 branch from cc266e4 to e0fd370 Compare January 25, 2022 06:52

lidavidm approved these changes May 19, 2022

View reviewed changes

lwhite1 suggested changes Nov 17, 2022

View reviewed changes

amol- closed this Mar 30, 2023

FiV0 mentioned this pull request Apr 8, 2024

GH-30866: [Java] fix SplitAndTransfer throws for (0,0) if vector empty #41066

Merged

asfimport mentioned this pull request Nov 16, 2022

[Java] SplitAndTransfer throws for (0,0) if vector empty #30866

Closed

lriggs mentioned this pull request Sep 3, 2024

Cherry pick fix for empty vectors and some additional build process fixes. dremio/arrow#81

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ARROW-15382: [Java] fix SplitAndTransfer throws for (0,0) if vector empty #12250

ARROW-15382: [Java] fix SplitAndTransfer throws for (0,0) if vector empty #12250

wzx140 commented Jan 25, 2022

github-actions bot commented Jan 25, 2022

wzx140 commented May 18, 2022

lidavidm commented May 18, 2022

lidavidm left a comment

pitrou commented Jun 7, 2022

lidavidm commented Nov 7, 2022

lwhite1 commented Nov 7, 2022

lwhite1 commented Nov 16, 2022

lwhite1 left a comment

amol- commented Mar 30, 2023

FiV0 commented Apr 2, 2024

ARROW-15382: [Java] fix SplitAndTransfer throws for (0,0) if vector empty #12250

ARROW-15382: [Java] fix SplitAndTransfer throws for (0,0) if vector empty #12250

Conversation

wzx140 commented Jan 25, 2022

github-actions bot commented Jan 25, 2022

wzx140 commented May 18, 2022

lidavidm commented May 18, 2022

lidavidm left a comment

Choose a reason for hiding this comment

pitrou commented Jun 7, 2022

lidavidm commented Nov 7, 2022

lwhite1 commented Nov 7, 2022

lwhite1 commented Nov 16, 2022

lwhite1 left a comment

Choose a reason for hiding this comment

amol- commented Mar 30, 2023

FiV0 commented Apr 2, 2024