jgoday commented on pull request #687:
URL: https://github.com/apache/arrow-datafusion/pull/687#issuecomment-876383651
@Jimexist I was trying to execute the integration tests manually,
executing test_psql_parity.py with the following sql
```
SELECT
c8,
LEAD(c8) OVER () next_c8,
LEAD(c8, 10, 10) OVER() next_10_c8,
LEAD(c8, 100, 10) OVER() next_overflow_c8,
LAG(c8) OVER() prev_c8,
LAG(c8, -2, 0) OVER() AS prev_2_c8,
LAG(c8, -200, 10) OVER() AS prev_overflow_c8
FROM test
ORDER BY c8;
```
There was some errors: type coercing problems between default_value and the
slice data_type
and between lead postgres (positive) and datafusion/shifting (negative)
arguments.
After fixed it in the latest commit, previous example works fine.
But if I try to test agains some custom partition
```sql
SELECT
LAG(c8) OVER(PARTITION BY c9 ORDER BY c9) AS prev_c8
FROM test
ORDER BY c8;
```
datafusion throws an error
```
Empty DataFrame
Columns: [Plan("Projections require unique expression names but the
expression \"LAG(#test.c8) AS prev_c8\" at position 4 and \"LAG(#test.c8)
PARTITION BY [#test.c9] ORDER BY [#test.c9 ASC NULLS FIRST] AS prev_c8\" at
position 7 have the same name. Consider aliasing (\"AS\") one of them.")]
```
I will try to look at this later today if I can ...
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
