Use String.concat to concatenate command line arguments

[saul]

While profiling design time builds I noticed that the Seq.fold here is responsible for most of the Fsc task's run time:

[KevinRansom]

@saul, I think the test failed because it expected an additional newline.

[cartermp]

I'm not sure how to read that chart, but isn't this saying that the 4.3s is coming from String.Concat anyways? That is, how much of this is due to the fold vs. how much if it is due to the String.Concat call made as a result of the + operator on strings?

[saul]

Yes all of the Seq.fold slowness is due to String.Concat (the BCL method). I replaced it with String.concat (the F# wrapper that passes to String.Join) as that uses StringBuilder under the hood. Sorry if my description was confusing!

[cartermp]

Thanks, looks good. I did a quick benchmark and it likely corroborates your own findings 🙂

// Learn more about F# at http://docs.microsoft.com/dotnet/fsharpopenSystemopenBenchmarkDotNet.AttributesopenBenchmarkDotNet.Running[<MemoryDiagnoser>]typeConcatBench()=letstrings=[for x in0..100-> sprintf "--commandlinearg:%d" x ][<Benchmark(Baseline=true)>]member_.SeqFold()=
strings
|> Seq.fold (fun acc f -> acc + f + Environment.NewLine)""|> ignore
[<Benchmark>]member_.FSharpCoreStringConcat()=
strings
|> String.concat Environment.NewLine
|> ignore
[<EntryPoint>]letmain argv =letsummary= BenchmarkRunner.Run<ConcatBench>()
printfn "%A" summary
0// return an integer exit code

.NET Framework:

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
SeqFold	17.498 μs	0.3491 μs	0.9136 μs	1.00	51.9714	-	-	213.04 KB
FSharpCoreStringConcat	2.837 μs	0.0562 μs	0.1161 μs	0.16	3.1357	-	-	12.86 KB

On .NET 5:

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
SeqFold	18.209 μs	0.3946 μs	1.1195 μs	1.00	51.8494	-	-	211.91 KB
FSharpCoreStringConcat	2.205 μs	0.0381 μs	0.0423 μs	0.12	1.0223	-	-	4.19 KB

[saul]

Really appreciate the quick turnaround, thanks both :)