Retain ordinal position of columns in get_columns_in_table method

[haylo75]

The result set for the default adapter's get_columns_in_table method returns columns unordered. In customizing dbt for my own use, I found it much easier to retain the original ordinal position. Debugging was simpler as I could compare directly to my database tool without mental friction.

This would be simple to implement, around line 188 in dbt/adapters/default.py

https://github.com/fishtown-analytics/dbt/blob/f039bb632df17b0cd2cf81b32b6d8366cb17b980/dbt/adapters/default.py#L188

Something like:
sql += "\norder by ordinal_position"

[drewbanin]

Good idea -- I think this should be pretty straightforward to add. Also need to figure out the analogues for snowflake and bigquery

[haylo75]

Cool - I should've mentioned that I didn't see this method overridden in the subclassed adapters. I assumed that, since the other PostgresAdapter derived adapters are using information_schema.columns, then the ordinal_position metacolumn was available in those platforms.

[haylo75]

yasss, thanks!