Clojurians Log v2

user=> (bench (simple-promise-chain-5))
amd64 Linux 5.2.9-arch1-1-ARCH 4 cpu(s)
OpenJDK 64-Bit Server VM 25.222-b05
Runtime arguments: -Dclojure.libfile=.cpcache/1379323951.libs
Evaluation count : 2330934 in 6 samples of 388489 calls.
      Execution time sample mean : 259.714484 ns
             Execution time mean : 259.613452 ns
Execution time sample std-deviation : 5.530111 ns
    Execution time std-deviation : 5.542603 ns
   Execution time lower quantile : 253.789804 ns ( 2.5%)
   Execution time upper quantile : 267.680303 ns (97.5%)
                   Overhead used : 1.535496 ns

user=> (bench (simple-completable-chain-5))
amd64 Linux 5.2.9-arch1-1-ARCH 4 cpu(s)
OpenJDK 64-Bit Server VM 25.222-b05
Runtime arguments: -Dclojure.libfile=.cpcache/1379323951.libs
Evaluation count : 1832040 in 6 samples of 305340 calls.
      Execution time sample mean : 265.049308 ns
             Execution time mean : 265.009442 ns
Execution time sample std-deviation : 3.174583 ns
    Execution time std-deviation : 3.400731 ns
   Execution time lower quantile : 261.017905 ns ( 2.5%)
   Execution time upper quantile : 268.803217 ns (97.5%)
                   Overhead used : 1.535496 ns

user=> (bench' (simple-promise-chain-5-raw))
Evaluation count : 4422288 in 6 samples of 737048 calls.
             Execution time mean : 137.411151 ns
    Execution time std-deviation : 1.440472 ns
   Execution time lower quantile : 135.616210 ns ( 2.5%)
   Execution time upper quantile : 139.088517 ns (97.5%)
                   Overhead used : 1.535496 ns
nil
user=> (bench' (simple-completable-chain-5-raw))
Evaluation count : 4780212 in 6 samples of 796702 calls.
             Execution time mean : 128.531062 ns
    Execution time std-deviation : 4.282302 ns
   Execution time lower quantile : 125.348313 ns ( 2.5%)
   Execution time upper quantile : 134.078223 ns (97.5%)
                   Overhead used : 1.535496 ns
nil

This one using plain then' without any sugaring (no realistic example, but show the real "overhead" of protocols and additional function calls)

hi!

yeah they both are quite similar, there are not millions of ways to wrap this stuff I guess. But while they are similar they differ in a few ways from what I saw when @ikitommi reminded me about promesa: no dyn vars, smaller api surface and I need manifold api compatibility (potentially, not using auspex in the real world for now), since our codebase is manifold heavy. But as @ikitommi said to me, in the end it's all completablefutures & co, so interoperable

(replying to @niwinz private msg here, felt relevant)

but before I consider using these I need a manifold/stream replacement, which is something I am considering/working-on atm (in spare/commute time)

@mpenet thanks for the answer, but with the lastest changes on the 4.0.0 there are very small differences i think.., and the missing api can be implemented... But on the end, you are right, they interoperate, I was just wondering if there are some colaboration 😊