It would be nice if the introspected output was deterministic. One way to do this is to sort the tables, columns, and other data before creating the final schema file.
I've made a quick fix for my use case, but I'm happy to push a fix if I'm pointed at the source code.
Also, IMO the snapshot shouldn't be saved for each migration but only used as input to creating the schema diff. It's redundant to keep a full snapshot of the schema for each version.
Hello!
It would be nice if the introspected output was deterministic. One way to do this is to sort the tables, columns, and other data before creating the final schema file.
I've made a quick fix for my use case, but I'm happy to push a fix if I'm pointed at the source code.
Thanks.