We should deprecate -release

[Lance Bachmeier]

On Wednesday, 17 July 2024 at 16:53:36 UTC, Richard (Rikki) 
Andrew Cattermole wrote:

> Dmd simply does not have the optimizations to get anywhere 
> close to a modern backend. It _cannot_ come close to winning 
> benchmarks.

This is the only point that matters in this discussion. DMD has a 
confusing switch that enables someone that doesn't use D to run a 
90% suboptimal benchmark instead of a 100% suboptimal benchmark.

DMD compiles quickly and it usually produces code that runs fast 
enough. It should never be used for benchmarking. If -release is 
how you detect someone doing benchmarking, it should fail to 
compile when using that flag.