Migration process plugin input and output

Migration process plugin input and output

Edit: 8f95381 implements 1, this is done.

Currently the migration process plugin always emits a list of destination ids. If you want a single id you need to run the extract plugin:

The roles source is a list of D6 role ids. Because it's a list and not a scalar, MigrateExecutable::processRow will iterate over it and feed every role id one by one to the migration process plugin. The output is array(d8_role_name) so we need to run extract to arrive to d8_role_name.

Although in an ideal world the plugin would only accept a list of source ids and emit a list of destination ids for the sake of the simple multiple handling, it is already accepting a scalar for a source id. However, as seen above it always answers with a list of destination ids. This assymetry is galling and confusing. There are two ways to restore symmetry:

1. When the migration process receives a scalar source id the output will be a scalar destination id (and throw an exception if there is more than destination id)
2. Add a wrap plugin, the counterpart of extract. Make the migration process plugin consistent and only accept an array of source ids. There's nothing else that would accept a scalar for source id values.

Both has pros and cons: 1. makes the migration configuration entity simpler but it makes source and destination ids inconsistent across the codebase. For those who only write the YAML this might not be that big of a problem but how many people will ever only write the YAML? When writing tests, it's a good idea to assert the correctness of the idmap for example. 2. makes the migration configuration entity longer to write for sure but it keeps source and destination ids consistently a list of ids.