Dolt Diff vs. Sqlite Diff Benchmark

bench.sh

#!/bin/bash

SCALE=100000
if [ ! -z "$1" ]; then
    SCALE="$1"
fi

# generate source data

cat << DOC > psv_gen.py
with open('$SCALE.psv', 'w+') as f:
	f.write('pk|c0|c1|c2\n')
	for i in range ($SCALE):
		s = '{}|{}|{}|{}\n'.format(i,i,i,i)
		f.write(s)
DOC

python3 psv_gen.py
rm psv_gen.py


# use source data to create and benchmark a dolt database

rm -rf .dolt
DOLT_DEFAULT_BIN_FORMAT="__DOLT_1__" dolt init > /dev/null
dolt table import -pk=pk -c difftbl "$SCALE.psv" > /dev/null
dolt add -A && dolt commit -m "created difftbl" > /dev/null
dolt sql -q "delete from difftbl where pk = 1234;" > /dev/null
dolt commit -am "create point diff" > /dev/null

echo "benchmarking dolt database (SCALE=$SCALE)"
time dolt sql -q "SELECT count(*) FROM dolt_commit_diff_difftbl WHERE to_commit=HASHOF('HEAD') AND from_commit=HASHOF('HEAD^')" > /dev/null
echo ""


# use source data to create and benchmark a sqlite database

rm diff.db  diff2.db
sqlite3 diff.db "create table difftbl (pk int primary key, c0 int, c1 int, c2 int)"
sqlite3 diff.db ".import $SCALE.psv difftbl"
cp diff.db diff2.db
sqlite3 diff2.db "delete from difftbl where pk = 1234"

echo "benchmarking sqlite3 database (SCALE=$SCALE)"
time sqldiff diff.db diff2.db > /dev/null
echo ""

# cleanup 

rm psv_gen.py
rm "$SCALE.psv"