When Getting “Help” is the Last Thing I Need
I've written a replacement to an exchange data feed and now I'm trying to work with a few groups to see if it'll work for them. The original intention was not to include them, so if it works for them, so much the better. Still, I've expected more than a little resistance because these are developers that have clearly stated that they do not want this new exchange feed - that the one they have is just fine, thank you. I understand their position - I'm new, they don't want to change. They fear what might happen, so it's easiest to simply drag their feet or poke holes in the project to say why they can't use it.
This is easily solved in one of three ways: fire people, force people, or give up. I'm not in a position to do any of these, but if I had my choice, I think I'd favor the ultimatum angle: force, then fire. This industry is paid far too well to put up with unnecessary crud from prima donnas thinking they know better.
Be that as it may... something struck me as quite odd, and made me lean towards the "firing" angle for one developer when he mentioned today that he took my code, replaced spinlocks with pthread locks and saw an amazing jump in performance. That makes no sense, as I moved from pthreads to spinlocks long ago for this exact reason. Still, I was willing to say I'd try it.
And I did.
Amazing. I got literally thousands of messages a second less using pthreads than using spinlocks - exactly what I expected. So much for the brilliant developer that thinks he's found the silver bullet to all problems.
Yup. For that developer, I'd give them a nice severance bonus and show them the door. Being reluctant is one thing. Being a bad developer is another.