Skip test_regexp_replace_null_pattern_fallback on Databricks [databricks] #4251
Conversation
…t Databricks Signed-off-by: Andy Grove <andygrove@nvidia.com>
|
if we are disabling this and closing the issue is it tracked by another issue to turn it back on? |
|
oh never mind, after looking at code looks like you are saying databricks optimizes that out, perhaps update description to say the same |
andygrove
changed the title
|
build |
| @@ -483,11 +483,14 @@ def test_regexp_replace(): | |||
| 'regexp_replace(a, "a|b|c", "A")'), | |||
| conf={'spark.rapids.sql.expression.RegExpReplace': 'true'}) | |||
|
|
|||
| @pytest.mark.xfail(condition=is_databricks_runtime(), | |||
There was a problem hiding this comment.
Do we want to xfail this or simply skip it? The point of xfailing is to track tests that we eventually expect to pass. I think skip is more appropriate for this one.
There was a problem hiding this comment.
Skipping it makes sense to me. Should I also update test_concat_ws_sql_col_sep_only_sep_specified in the same file to skip instead of xfail? (that's where I copied the pattern from).
There was a problem hiding this comment.
Yes, any test that we are never expecting to pass because it's an invalid premise for testing in a particular environment should be skipped rather than xfailed. We should reserve xfail for tests we think should pass at some point, as they essentially serve as TODO's in the test output.
andygrove
changed the title
|
build |
Signed-off-by: Andy Grove andygrove@nvidia.com
Closes #4243