GH-47648: [Archery][Integration] More granularity in JSON test cases#47649
GH-47648: [Archery][Integration] More granularity in JSON test cases#47649pitrou merged 1 commit intoapache:mainfrom
Conversation
|
@Alex-PLACET does this look ok? |
|
@pitrou perfect 👌 |
|
@alamb @tustvold @zeroshade Is it ok to you? |
paleolimbot
left a comment
There was a problem hiding this comment.
Sounds good!
It was already not 1:1, but I found it confusing when I started out that the files in arrow-testing didn't quite align with the ones generated by Archery and I suppose this further widens that gap? The golden files in arrow-testing are easier to use than a full-on integration test for those getting started.
For the old files, yes, but the plan is to add a set of more recent "golden" files generated with Arrow 21: #47635 |
16922e5 to
1b7808c
Compare
|
@zeroshade How can we make these errors go away on the integration tests? They've been happening randomly for quite some time. |
|
Ok, it should be fixed by #47674 |
…cases Separate binary columns from other primitives in the integration test cases, to make life easier when writing a new Arrow implementation.
1b7808c to
c755612
Compare
|
After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit eb9d519. There were no benchmark performance regressions. 🎉 The full Conbench report has more details. It also includes information about 3 possible false positives for unstable benchmarks that are known to sometimes produce them. |
…cases (apache#47649) ### Rationale for this change Separate binary columns from other primitives in the integration test cases, to make life easier when writing a new Arrow implementation. ### Are these changes tested? Yes, by integration CI test. ### Are there any user-facing changes? No, only for implementers who would use the generated files directly. * GitHub Issue: apache#47648 Authored-by: Antoine Pitrou <antoine@python.org> Signed-off-by: Antoine Pitrou <antoine@python.org>
Rationale for this change
Separate binary columns from other primitives in the integration test cases, to make life easier when writing a new Arrow implementation.
Are these changes tested?
Yes, by integration CI test.
Are there any user-facing changes?
No, only for implementers who would use the generated files directly.