Up to Main Index Up to Journal for November, 2022
JOURNAL FOR WEDNESDAY 30TH NOVEMBER, 2022
______________________________________________________________________________
SUBJECT: Not a good month for the journal…
DATE: Wed 30 Nov 20:15:25 GMT 2022
Only two posts this month, including this one :(
Work has been really busy and Mere has been progressing slowly. The number of
tests for operators and built-ins has risen from 2700 to 2980 asserts. The
quality of the tests has improved a lot and as a result many bugs fixed.
My ugly Bash script for testing has been replaced. Tests are now run using
Go’s standard testing framework:
>go test ./mere/...
ok code.wolfmud.org/mere.git/mere 0.043s
>
This also means I can generate coverage reports — which I’ve been using to
improve the tests:
>go test -coverprofile cover.out ./mere/...
ok code.wolfmud.org/mere.git/mere 0.058s coverage: 78.0% of statements
>go tool cover -html=cover.out
# This opens the code coverage report in a web browser…
Tests can be benchmarked and memory/CPU profiles are easier to obtain. For
anyone interested, a sample for the tests[1] and individual operators[2] is
included at the end. Note that the individual operator tests are still a work
in progress.
The tests are still written in Mere itself, but compiled and run under the
test framework:
func TestMere(t *testing.T) {
tests, _ := filepath.Glob(filepath.Join("..", "testdata", "*.mr"))
for _, name := range tests {
t.Run(name, func(t *testing.T) {
src, _ := os.ReadFile(name)
i := Compile(string(src))
if i.Execute() != 0 {
t.Fatal("")
}
})
}
}
For benchmarks, the tests are compiled once and executed multiple times. So
the times and allocations are only for executing the code, not compiling it:
func BenchmarkMere(b *testing.B) {
tests, _ := filepath.Glob(filepath.Join("..", "testdata", "*.mr"))
for _, name := range tests {
src, _ := os.ReadFile(name)
I := Compile(string(src))
b.Run(filepath.Base(name), func(b *testing.B) {
for i := 0; i < b.N; i++ {
if I.Copy().Execute() != 0 {
b.Fatal("")
}
}
})
}
}
The “compile once and run multiple” times is how I imagine WolfMUD will use
Mere. It’s already possible to run multiple instances of a compiled program. I
see Mere programs as small snippets, used when defining simple static rules in
zone files is not enough.
I’ve finished work for the day, written this and *poof* my evening has gone :(
At least the night is still young, perfect time for coding ;)
--
Diddymus
[1] Sample benchmark run of the tests:
>go test -bench="Mere" -benchmem ./mere/...
goos: linux
goarch: amd64
pkg: code.wolfmud.org/mere.git/mere
cpu: Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz
B…/b_assert.mr-4 22419 52182 ns/op 16405 B/op 138 allocs/op
B…/b_bool_array.mr-4 13854 88076 ns/op 20855 B/op 218 allocs/op
B…/b_bool_map.mr-4 5790 214657 ns/op 53233 B/op 496 allocs/op
B…/b_delete.mr-4 5842 192327 ns/op 42835 B/op 491 allocs/op
B…/b_exists.mr-4 14170 89446 ns/op 22407 B/op 216 allocs/op
B…/b_float_array.mr-4 10000 100019 ns/op 29358 B/op 226 allocs/op
B…/b_gosub.mr-4 135493 8859 ns/op 2447 B/op 42 allocs/op
B…/b_goto.mr-4 176642 8164 ns/op 2327 B/op 38 allocs/op
B…/b_int_array.mr-4 10000 102459 ns/op 29183 B/op 221 allocs/op
B…/b_int_map.mr-4 7322 162947 ns/op 39545 B/op 364 allocs/op
B…/b_keys.mr-4 49760 23224 ns/op 6897 B/op 92 allocs/op
B…/b_len.mr-4 30273 41743 ns/op 13609 B/op 110 allocs/op
B…/b_literal.mr-4 4849 282871 ns/op 61154 B/op 1151 allocs/op
B…/b_sort.mr-4 17623 64587 ns/op 15585 B/op 130 allocs/op
B…/b_sprint.mr-4 15076 78952 ns/op 17465 B/op 405 allocs/op
B…/b_string_array.mr-4 10000 106694 ns/op 38255 B/op 204 allocs/op
B…/b_substr.mr-4 106426 11535 ns/op 2684 B/op 57 allocs/op
B…/issues.mr-4 20940 57074 ns/op 6896 B/op 117 allocs/op
B…/op_add.mr-4 28041 43080 ns/op 9151 B/op 135 allocs/op
B…/op_and.mr-4 82416 16425 ns/op 3430 B/op 35 allocs/op
B…/op_assign.mr-4 3574 324021 ns/op 66738 B/op 574 allocs/op
B…/op_bool.mr-4 42499 31609 ns/op 6544 B/op 70 allocs/op
B…/op_dec.mr-4 67170 21144 ns/op 5381 B/op 67 allocs/op
B…/op_div.mr-4 58701 18942 ns/op 3780 B/op 57 allocs/op
B…/op_eq.mr-4 9580 129701 ns/op 24062 B/op 242 allocs/op
B…/op_float.mr-4 20192 57297 ns/op 7606 B/op 101 allocs/op
B…/op_ge.mr-4 22231 56760 ns/op 8419 B/op 92 allocs/op
B…/op_gt.mr-4 22413 57269 ns/op 8419 B/op 92 allocs/op
B…/op_inc.mr-4 54649 21648 ns/op 5381 B/op 67 allocs/op
B…/op_index.mr-4 14078 90335 ns/op 19102 B/op 217 allocs/op
B…/op_index_assign.mr-4 1579 819121 ns/op 108057 B/op 1121 allocs/op
B…/op_int.mr-4 44778 28096 ns/op 6729 B/op 76 allocs/op
B…/op_land.mr-4 84967 15305 ns/op 2768 B/op 34 allocs/op
B…/op_le.mr-4 21312 55594 ns/op 8417 B/op 92 allocs/op
B…/op_lor.mr-4 77998 17538 ns/op 3184 B/op 35 allocs/op
B…/op_lt.mr-4 21922 57425 ns/op 8417 B/op 92 allocs/op
B…/op_mod.mr-4 19261 61334 ns/op 8130 B/op 147 allocs/op
B…/op_mult.mr-4 54520 23925 ns/op 4558 B/op 76 allocs/op
B…/op_mult_assign.mr-4 257280 5560 ns/op 1765 B/op 23 allocs/op
B…/op_ne.mr-4 13566 88211 ns/op 16562 B/op 137 allocs/op
B…/op_neg.mr-4 119104 12198 ns/op 2980 B/op 54 allocs/op
B…/op_not.mr-4 154328 9547 ns/op 2768 B/op 34 allocs/op
B…/op_or.mr-4 89485 15911 ns/op 3427 B/op 35 allocs/op
B…/op_shift_left.mr-4 93109 15644 ns/op 3809 B/op 46 allocs/op
B…/op_shift_right.mr-4 113055 12560 ns/op 3160 B/op 36 allocs/op
B…/op_spaceship.mr-4 16782 71610 ns/op 11064 B/op 190 allocs/op
B…/op_string.mr-4 21387 59520 ns/op 12164 B/op 226 allocs/op
B…/op_sub.mr-4 45928 29858 ns/op 6128 B/op 107 allocs/op
B…/op_type.mr-4 47056 29723 ns/op 8046 B/op 117 allocs/op
B…/op_xor.mr-4 68089 16569 ns/op 3426 B/op 35 allocs/op
PASS
ok code.wolfmud.org/mere.git/mere 78.608s
[2] Sample benchmark run of individual operators:
>go test -bench="Ops" -benchmem ./mere/...
goos: linux
goarch: amd64
pkg: code.wolfmud.org/mere.git/mere
cpu: Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz
BenchmarkOps/b-4 724317 1963 ns/op 1200 B/op 18 allocs/op
BenchmarkOps/b=-4 503775 2368 ns/op 1248 B/op 19 allocs/op
BenchmarkOps/b=b-4 589674 2378 ns/op 1248 B/op 19 allocs/op
BenchmarkOps/ba-4 708888 1944 ns/op 1392 B/op 18 allocs/op
BenchmarkOps/bm-4 590388 1934 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/f-4 623192 1925 ns/op 1248 B/op 18 allocs/op
BenchmarkOps/f=-4 528535 2215 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/f=Af+Af-4 452374 2531 ns/op 1400 B/op 21 allocs/op
BenchmarkOps/f=f-4 531918 2393 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/f=f+f-4 452587 2853 ns/op 1400 B/op 21 allocs/op
BenchmarkOps/fa-4 586860 1898 ns/op 1392 B/op 18 allocs/op
BenchmarkOps/fm-4 604845 2060 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/i-4 725466 1875 ns/op 1248 B/op 18 allocs/op
BenchmarkOps/i=-4 572564 2207 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/i=Ai+Ai-4 543379 2586 ns/op 1392 B/op 20 allocs/op
BenchmarkOps/i=i-4 594530 2434 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/i=i+i-4 483459 2684 ns/op 1392 B/op 20 allocs/op
BenchmarkOps/ia-4 609939 1977 ns/op 1392 B/op 18 allocs/op
BenchmarkOps/im-4 574332 2147 ns/op 1296 B/op 19 allocs/op
BenchmarkOps/s-4 688345 1917 ns/op 1312 B/op 18 allocs/op
BenchmarkOps/s=-4 591752 2403 ns/op 1360 B/op 19 allocs/op
BenchmarkOps/s=As+As-4 525708 2745 ns/op 1474 B/op 22 allocs/op
BenchmarkOps/s=s-4 506758 2560 ns/op 1360 B/op 19 allocs/op
BenchmarkOps/s=s+s-4 489769 2903 ns/op 1474 B/op 22 allocs/op
BenchmarkOps/sa-4 562858 2046 ns/op 1392 B/op 18 allocs/op
BenchmarkOps/sm-4 687019 2017 ns/op 1296 B/op 19 allocs/op
PASS
ok code.wolfmud.org/mere.git/mere 35.197s
Up to Main Index Up to Journal for November, 2022