I want to write the following process:

• Register a record in Table A (primary key is ‘hoge’)
• Register a record in Table B (set ‘hoge’ as the foreign key)

When inserting into a database, it’s possible to generate IDs (sequential or UUID) on the database side.

• Method A: Generate IDs in the program and register them in the database.

  1. Generate the ID (‘hoge’) in the program.
  2. Register the record in Table A (set ‘hoge’ as the primary key).
  3. Register the record in Table B (set ‘hoge’ as the foreign key).

  This feels natural to me.

• Method B: Let the database generate IDs.

  1. Register a record in Table A (with ‘hoge’ as the primary key).
  2. Retrieve the primary key ‘hoge’ from Table A.
  3. Register a record in Table B (set ‘hoge’ obtained in step 2 as the foreign key).

  This results in extra database communication to retrieve the primary key ‘hoge’.

I would like to hear your opinions (I prefer Method A).