In his characteristic style Rico talks about what it would take the “beat” (from a perf point of view) the CLR implementation of generics with hand authored code. In addition it reiterates some key principles on how NOT to make technology choices. I’ll add a couple to that:
My mom uses feature X…
I saw Don Box talk about feature X in his underwear…
I hear Chris Anderson talks about feature X for 3 hours with zero slides
Dr. Dobbs did a cover article on feature X….
This is true even if X=”Managed Code”. I mean, really, as much of a managed code zealot as I am, even *I* would say don’t use managed code if the cost of managed code does not have good cost/value proposition (deal) in your scenario. I certainly hope the productivity, scalability, deployment capabilities, security, 32-64 bit interoperability, developer support and community, and great IDE all make managed code worth it for your next project. But if they don’t, by all means don’t use it.
What are your bad reasons for choosing a technology? Where have you opted to not use a feature because the “deal” wasn’t good?