| 1 | #!/usr/bin/env bash
 | 
| 2 | #
 | 
| 3 | # Usage:
 | 
| 4 | #   benchamrks/builtin-io.sh <function name>
 | 
| 5 | 
 | 
| 6 | set -o nounset
 | 
| 7 | set -o pipefail
 | 
| 8 | set -o errexit
 | 
| 9 | 
 | 
| 10 | readonly BIG=_tmp/1m_lines.txt
 | 
| 11 | 
 | 
| 12 | setup() {
 | 
| 13 |   seq 1000000 > $BIG
 | 
| 14 | 
 | 
| 15 | }
 | 
| 16 | 
 | 
| 17 | # 25 ms
 | 
| 18 | wc-big() {
 | 
| 19 |   time wc -l $BIG
 | 
| 20 | }
 | 
| 21 | 
 | 
| 22 | # bash takes 156 ms here!  Significantly slower than 'wc'.
 | 
| 23 | # bin/osh in Python takes over 5 seconds!
 | 
| 24 | #
 | 
| 25 | # TODO:
 | 
| 26 | # - Make sure bin/osh in C++ is reasonably fast.
 | 
| 27 | # - Make sure a loop with read --line is reasonably fast.
 | 
| 28 | 
 | 
| 29 | mapfile-big() {
 | 
| 30 |   time mapfile < $BIG
 | 
| 31 |   echo ${#MAPFILE[@]}  # verify length
 | 
| 32 | }
 | 
| 33 | 
 | 
| 34 | # Hm this isn't that fast either, about 100 ms.
 | 
| 35 | python-big() {
 | 
| 36 |   time python -S -c '
 | 
| 37 | import sys
 | 
| 38 | i = 0
 | 
| 39 | for line in sys.stdin:
 | 
| 40 |   i += 1
 | 
| 41 | print(i)
 | 
| 42 | ' < $BIG
 | 
| 43 | }
 | 
| 44 | 
 | 
| 45 | bash-syscall() {
 | 
| 46 |   # Shows that there are tons of read(0, 1) calls!
 | 
| 47 |   seq 20 | strace -e read -- bash -c 'mapfile'
 | 
| 48 | }
 | 
| 49 | 
 | 
| 50 | python-syscall() {
 | 
| 51 |   # Does read(0, 4096).  A saner way to read files
 | 
| 52 |   seq 20 | strace -e read -- python -c '
 | 
| 53 | import sys
 | 
| 54 | for line in sys.stdin:
 | 
| 55 |   print(line)
 | 
| 56 | '
 | 
| 57 | }
 | 
| 58 | 
 | 
| 59 | 
 | 
| 60 | "$@"
 |