Open AndBobsYourUncle opened 5 years ago
@AndBobsYourUncle Hi there. Is this still something you're interested in having? If so, is it still an issue?
Could you expand a bit further on the use case including perhaps a positive and negative example where you want the ignored_tables behavior to apply, and one where you don't? For example you might choose to write this as a little bit of documentation showing how to use it. That might help in understanding the use case as well.
If there's no response after maybe a month, we'll likely close this issue out.
Thanks!
This adds the ability to ignore certain tables when doing column reordering.
Having this feature is particularly useful when there are Postgres DB functions (or other such things) that rely on implied column names from the ordering of the values in the query.
In particular, there is a case where a Postgres function in our project has an INSERT statement like this:
INSERT INTO some_table VALUES(1, 'some_data', 2);
The columns are implied from the ordering, so if the ordering changes, this breaks the assumption. In CircleCI, the
structure.sql
is used to initialize the test DB, so integrated testing fails with the new column ordering.