Rationale is following:
If you run pgloader with drop schema, whatever schema modifications are done in BEFORE LOAD are gone after the schema is dropped.
If you put them in AFTER SCHEMA CREATE, they will have no effect on things happening during the schema creation.
For example, if you need to create an extension in the schema that makes a new data type, and include that datatype into a casting rule, the extension has to be installed into the schema prior to creation of tables, but after the schema has been created.
You can of course hardcode your schema deletion in BEFORE LOAD itself, but that seem to obviate the point of automating this process.
Rationale is following:
If you run pgloader with drop schema, whatever schema modifications are done in BEFORE LOAD are gone after the schema is dropped.
If you put them in AFTER SCHEMA CREATE, they will have no effect on things happening during the schema creation.
For example, if you need to create an extension in the schema that makes a new data type, and include that datatype into a casting rule, the extension has to be installed into the schema prior to creation of tables, but after the schema has been created.
You can of course hardcode your schema deletion in BEFORE LOAD itself, but that seem to obviate the point of automating this process.