Refactor whole structures because GradualPerfomanceAttributes cannot be shared between threads properly. So now we have clear distinction between what structs can be shared between threads and which is not
Switch from HashMap to simple Vec when storing clients
Drop crossbeam-channel dependency since we don't need it anymore
For all HTTP request will be spawned separate async task (using smol runtime)
Sadly binary size gonna grow from 1.9MB to 2.4MB (linux)
GradualPerfomanceAttributescannot be shared between threads properly. So now we have clear distinction between what structs can be shared between threads and which is notHashMapto simpleVecwhen storing clientscrossbeam-channeldependency since we don't need it anymoreFor all HTTP request will be spawned separate async task (using
smolruntime)Sadly binary size gonna grow from 1.9MB to 2.4MB (linux)