News for hackers. Stuff that matters.
Re:PostgreSQL and UUID as Primary Key by bruce511 on Saturday July 06, 2024 @12:49AM (#40888227)
Yes, sure, it leaks some information - but to be fair printing an invoice also leaks information.

For me the priority is security. If I get a link (visible or invisible) that contains a numeric ID, there's the possibility to tweak that link with another number.

Ideally, the server treats that number as suspect. Every. Single . Time. In practice I only need one developer to miss the check in one place and I have a serious security issue. Like leaking my entire customer list serious. This has happened multiple times -that we know of-.

Yes obscurity is not security. But sequential numbers are insecure by design, and it takes a lot of well-maintained walls to protect them. One crack leaves them very vulnerable.

Re:PostgreSQL and UUID as Primary Key by sgarland on Saturday July 06, 2024 @09:15AM (#40890269)
Call me naïve, but surely you can have fuzzing tests in CI?

I’m also going to use this as yet another example of why getting rid of QA in favor of Ship It Now was a bad idea.

Re:PostgreSQL and UUID as Primary Key by bruce511 on Saturday July 06, 2024 @02:51PM (#40892349)
Relying on tests to discover the tortuous path to most security exploits would not be my recommendation.