regex_column_names users feedback

[Pablo Langa Blanco <so...@gmail.com>]

Hi Spark devs & users,

I’m writing to get some feedback from the users of the regex_column_names
feature (spark.sql.parser.quotedRegexColumnNames) (
https://spark.apache.org/docs/latest/sql-ref-syntax-qry-select.html)

Now, some queries like  SELECT `col_.*`/col_b FROM (SELECT 3 AS col_a, 1 as
col_b) are not allowed but in some cases, when the regular expression
resolves to only one column, it could be resolved.

For example:

   -

   SELECT `col_.*`/exp FROM (SELECT 3 AS col_a, 1 as exp) --> Could be
   resolved to SELECT col_a/exp FROM (SELECT 3 AS col_a, 1 as exp)
   -

   SELECT `col_a`/exp FROM (SELECT 3 AS col_a, 1 as col_b) -->  Could be
   resolved to SELECT col_a/exp FROM (SELECT 3 AS col_a, 1 as exp)


Does it make sense for you? Or it’s confusing and it’s preferable to fail?

Thanks

Regards