I just had a Jr dev do this to me. Came in, so bold, and said our patterns were stupid. Told him no, while there are things to improve it’s not worth the effort to rewrite hundreds of classes over a personal preference.
Of course when he inevitably brought it up again, I told him okay, schedule with our architect, that’s how things work after all for major rewrites. Man, was that meeting entertaining.
Kids, we know it could be better. But we also know things like level of effort and value. Your CS330 patterns are really cool, we know them too. But we aren’t going to build them.
Oh god this was a trigger. Had one guy just casually demand that we change our middleware layers out because it would be “more performant”. Our middleware that was rock solid, never had any issues, and no one ever noticed. He brought up how it would shave so much time, like maybe a ms or 2 when I read through what he proposed. Sure, let’s just throw out what’s been working for years without issue for a new, untested version that will take weeks to dev and test for a payout of 2 ms per request.
“I could rewrite this in a week!”
~ junior dev, 3 months ago
I just had a Jr dev do this to me. Came in, so bold, and said our patterns were stupid. Told him no, while there are things to improve it’s not worth the effort to rewrite hundreds of classes over a personal preference.
Of course when he inevitably brought it up again, I told him okay, schedule with our architect, that’s how things work after all for major rewrites. Man, was that meeting entertaining.
Kids, we know it could be better. But we also know things like level of effort and value. Your CS330 patterns are really cool, we know them too. But we aren’t going to build them.
No no you see if we ditch QT we can create our own widgets in open GL it would be less overhead trust me bru
Oh god this was a trigger. Had one guy just casually demand that we change our middleware layers out because it would be “more performant”. Our middleware that was rock solid, never had any issues, and no one ever noticed. He brought up how it would shave so much time, like maybe a ms or 2 when I read through what he proposed. Sure, let’s just throw out what’s been working for years without issue for a new, untested version that will take weeks to dev and test for a payout of 2 ms per request.
3 years later, hands unfinished work to junior dev after the system they re-invented has also become obsolete