Question

I’m not sure how to search for this, so here it is.

My own terms for this are “synchronized” and “deviated” database schemas. If there’s an official term for all this I’d like to know it.

This might seem a bit stupid, but I’ve been stuck on which direction I want to go. I’m new to creating database schemas. I’m using SQLite on a Java application, and using a single .db file to contain different categories of information in different tables. So the first table contains accountID/username/password; and maybe the next table would be accountID/firstname/lastname/email. I plan on adding additional tables in the future as new features are implemented.

The decision I’m having a problem making here is… if a user makes a basic account, but doesn’t fill out anything else– should I make a blank row for those other tables? Or should I not add a row, and make the row when they do enter data for those tables(which could be months later)? I hope this makes sense.

What are the pros/cons? Implementation/programming? Efficiency? etc.

Thanks

Answer 1

IMO, all the single-valued user attributes should reside in a single table, i.e. users table would contain the following columns:

• accountId
• userName
• password
• firstName
• lastName
• email

When onboarding a user, you can specify each attribute as either mandatory or optional. Optional attributes will be NULLable columns in the database. When the user specifies values to these attributes, you would just update them in the table then.

Going forward, while adding new features, if you happen to come across a single-value attribute of a user, you would need to alter the users table and add that column with default value as null.

If the frequency of adding new attributes to user is too high, then you can move towards an attribute table model where each row will store one additional attribute of the user. There would be a one-to-many relationship between the users table and this user_attrs table. HTH.