@Ocramius I've re-added the @pure annotations, so static analysis is now failing (as expected, I guess). I could use some direction on what's needed to fix the failures. Thanks!

👋 Hello! I've spent some time on this branch, and here's my current thinking, along with an approach.

• First principles: As we all know, a pure function is a function that will return the same value every time it's called with the same inputs, and has no side effects in the codebase.
• First principles: ramsey/uuid also features a factory subsystem that allows for pluggable uuid generation, which is used internally for automated testing purposes. In it's current state, this factory subsystem uses static variables to change how it creates UUID's, which by definition makes anything it touches an impure function. With normal usage, this subsystem doesn't change, and based on the codebases's internal comments, can be considered pure with normal usage.
• Some of the methods we label as @pure\@psalm-pure are legitimately pure functions, and should keep the @pure label. For example, the comment on BinaryUtils from the original PR is a legitimate pure function. It's also passing the current phpstan scan as is.
• The code where phpstan is currently failing are largely methods either directly or indirectly touching this factory system. In earlier versions of the codebase, there were @psalm-ignore directives that suppressed errors. We should add php-stan equivalent ignore directives with comparable notes.

I'm planning on working through the error list this morning, but wanted to drop this into the PR thread to make sure that folks were on board with this approach before getting too far into it.

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.11%. Comparing base (dbc8104) to head (198a34a).

✅ All tests successful. No failed tests found.

@@             Coverage Diff              @@
##                4.x     #603      +/-   ##
============================================
+ Coverage     95.06%   95.11%   +0.04%     
  Complexity      574      574              
============================================
  Files            70       70              
  Lines          1622     1637      +15     
============================================
+ Hits           1542     1557      +15     
  Misses           80       80              

Thanks so much for your help, @fillerwriter!

@Ocramius, I'd love for you to take a final glance at this before I merge it into the 4.x branch. We're now adding @pure in some places where it wasn't before, and I want to make sure this doesn't also create a BC break.

If it's considered new, backward-compatible functionality, then all is well, and I'll tag a 4.x.0 release. Otherwise, I want to remove the annotations that cause BC breaks for 4.x, while keeping everything for the 5.x branch.

@fillerwriter, I addressed your concerns about the PHPStan errors in the tests here: 198a34a

I'm still not sure why it only just now started complaining about that.

Nice! Re: PHPStan errors in tests, that's a nice trick! I didn't know you could filter errors based on the folder.

I didn't know you could filter errors based on the folder.

I didn't, either, but I suspected you could, so I figured that out tonight. 🙂

@ramsey fwiw, @fillerwriter's analysis is correct, and the suppressions correspond to my initial design too (which is also what highlighted the entire factory problematic under the hood)