Re:What “Work” Looks Like
by uglygoblin on Tuesday October 25, 2022 @12:05PM
(#33332444)
I empathize deeply with you. I have been dealing with a system of "infinite configurability" for the last couple years and it's playing out the same way. The best I've been able to do is slowly codify the most problematic corners and put guard rails as many places as possible.
Sometimes I want to put my hands up in defeat and leave but there is also a satisfaction and growth from simplifying the Rube Goldberg machine.
It still sucks though.
"Code coverage" and code that does everything and thus nothing are a potent cocktail as well. You can have 90% 'coverage' on a file and still only be covering 5% of the actual code paths (that's unfortunately not hyperbole, but a real number from a real analysis I did. When the numbers disagree that much, people tend to ignore the terrifying one)
> But that was after having tried several other solutions,
I should clarify that I didn't mean concurrent or consecutive here, but across several different projects. Different preconditions can sink a strategy. It's hard to get any signal when you apply multiple unsuccessful strategies to the same 'experiment'.
I’m working at a place like that now and it’s been an eye roller. After ten years on the project one of the problem people has finally seen this as a problem (though I haven’t heard him admit that he contributes). He’s looking at code now written by two people who have been copying his code style for years, and suddenly feeling the pain of it. It’s great he’s growing, but he’s over forty and should have learned this years ago. I blame a mix of things including staying at one place for far too long.
And I keep telling him this code used to be worse, but I’ve been chipping away at it for some time.