Re: [Help-bash] Why `$(<` is slow?

[Eli Schwartz]

On 8/29/19 9:39 PM, Peng Yu wrote:
> I got the following runtime. `read` is the fastest (the last time statement).
> 
> But `$(<` is doing the same thing (the 4th time statement), shouldn't
> it be as quick as read?
> 
> The assignment `=` takes little time (the 3rd time statement), so that
> I think the most runtime is spent in `$(<` in the 4th time statement.
> 
> `$(<` is just a little faster than `$(cat`. Is I guess `$(<` involves
> a subshell? 

Yes.

As you suspected, you already knew the answer.

> But shouldn't subshell be optimized away in such a case?

Why?

> Thanks.
> 
> $  ./main.sh
> time for((i=0;i<1000;++i)); do ( cat "$tmpfile" ) done
> 
> real  0m1.112s
> user  0m0.402s
> sys   0m0.701s
> time for((i=0;i<1000;++i)); do x=$(cat "$tmpfile"); done
> 
> real  0m1.218s
> user  0m0.459s
> sys   0m0.720s
> time for((i=0;i<1000;++i)); do x=1; done
> 
> real  0m0.009s
> user  0m0.008s
> sys   0m0.001s
> time for((i=0;i<1000;++i)); do x=$(< "$tmpfile"); done
> 
> real  0m1.169s
> user  0m0.407s
> sys   0m0.727s
> time for((i=0;i<1000;++i)); do read -d '' -r x < "$tmpfile"; done
> 
> real  0m0.048s
> user  0m0.027s
> sys   0m0.020s
> $ cat main.sh
> #!/usr/bin/env bash
> # vim: set noexpandtab tabstop=2:
> 
> tmpfile=$(mktemp)
> enable -f cat cat
> set -v
> time for((i=0;i<1000;++i)); do ( cat "$tmpfile" ) done
> time for((i=0;i<1000;++i)); do x=$(cat "$tmpfile"); done
> time for((i=0;i<1000;++i)); do x=1; done
> time for((i=0;i<1000;++i)); do x=$(< "$tmpfile"); done
> time for((i=0;i<1000;++i)); do read -d '' -r x < "$tmpfile"; done
> 


-- 
Eli Schwartz
Bug Wrangler and Trusted User

signature.asc
Description: OpenPGP digital signature