[malinens]

if you keep using sqlite for tests it forces Your database logic to be universal. You could confidently switch to other db like mysql any time.

[mumblemumble]

It also means that you have to stick to the lowest common denominator, which is approximately equivalent to the state of the art as of a quarter century ago.

Every benefit has a cost. The benefit doesn't always justify the cost.

edit: To add to that - I've seen more than a couple major database engine migrations in my day, so it's not to say that that isn't a concern. But none of them has ever been from one SQL RDBMS to another. More common is migrating among different classes of database. MySQL to Mongo, Oracle to BigTable, Couch to Cassandra, something like that. MS Access to MS SQL Server a couple times, but even those are different enough that it was never going to be as simple as changing the connection string and having a carefree life.

The speculative future proofing that you do almost never manages to work for the future you end up actually living.

[taffer]

This also means that you have to stick to the lowest common denominator between all databases.

[harrisonjackson]

This was specifically why we moved away from sqlite. We wanted to take advantage of features postgres offers that are not universal.