huh, that's a good catch. I'm not sure about the proposed solution though. on production the currently committed schema is probably correct since the migrations were applied in merge order. as long as we query by column name this order mismatch shouldn't matter too much, but if we query by column order this might be problematic.

@weiznich do you have any suggestions for situations like this? :)

There is a column_sorting option for diesel.toml files that could be used to address this. At currently has these options. I really need to get around documenting all the missing options on the web page…

That written: Diesel always requests columns by name in your select statement, so a different order in the table! definition shouldn't cause any issues as long as you don't rely on deserializing the data with #[derive(Queryable)] only (so no Selectable) and the fields in question all have the same type or as long as you don't write any custom select via QueryFragment and use * there. If you using Selectable (or the newer HasQuery derive) it should never be an issue.