ngervasi wrote:UUIDs are sexy but are very difficult to handle during debugging (integer pks easily show records creation order and are easy to spot when debugging), another concern is about performance if you have very large tables.
David, are you experiencing any performance issue on queries? Especially joins?
All total bunk
In what cases is the PK type a query performance consideration? If you know this answer, why are you developing in Servoy? And I had a developer not long ago state that they were using integers instead of UUIDs because they were easier to remember and type when debugging. This is not a valid reason!
SaaS, sharing data, syncing offline data, distributed data, multiple applications on same data, security, no maintaining uniqueness overhead, no scaling issues, multi-threaded ok, mobile friendly, cloud friendly, database vendor agnostic, unique across tables...these are valid reasons.
Auto-increment integers didn't have to deal with all these situations back in the day. And nowadays with compression implemented by many databases, UUID space cost is negligible.
Many of the new db's in the past handful of years don't even have auto-increment integer PKs as an option.