article about speed

[eitsupi]

I am curious about the speed compared to the semver package with a parser written in C++, so some documentation showing this would be great.

[eitsupi]

More than 1000 times slower!

version_chrs <- sprintf(
  "1.2.3-foo.%s",
  1:100
)

bench::mark(
  semver = semver::parse_version(version_chrs),
  smvr = smvr::parse_semver(version_chrs),
  min_iterations = 10,
  check = FALSE
)
#> Warning: Some expressions had a GC in every iteration; so filtering is
#> disabled.
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 semver        153µs    196µs   3809.      5.53MB     8.00
#> 2 smvr          517ms    567ms      1.75     2.1MB    14.7

^{Created on 2025-06-28 with reprex v2.1.1}

[eitsupi]

#15 makes this faster.

version_chrs <- sprintf(
  "1.2.3-foo.%s",
  1:100
)

bench::mark(
  semver = semver::parse_version(version_chrs),
  smvr = smvr::parse_semver(version_chrs),
  min_iterations = 10,
  check = FALSE
)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 semver     160.33µs  198.8µs    4300.     5.53MB     9.28
#> 2 smvr         8.37ms   10.9ms      91.0  970.15KB    17.2

^{Created on 2025-06-28 with reprex v2.1.1}

[eitsupi]

Interestingly, smvr seems to be faster when sorting after parsing.
(In other words, semver sorting is quite slow.)

version_chrs <- sprintf(
  "1.2.3-foo.%s",
  1:100
)

bench::mark(
  semver = semver::parse_version(version_chrs) |> sort(),
  smvr = smvr::parse_semver(version_chrs) |> sort(),
  min_iterations = 10,
  check = FALSE
)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 semver      17.27ms   18.3ms      53.1    5.62MB     37.9
#> 2 smvr         7.81ms    8.8ms     110.   984.45KB     20.0

^{Created on 2025-06-30 with reprex v2.1.1}

set.seed(42)

version_chrs <- sprintf(
  "1.2.3-foo.%s",
  100:1
) |>
  sample()

bench::mark(
  semver = semver::parse_version(version_chrs) |> sort(),
  smvr = smvr::parse_semver(version_chrs) |> sort(),
  min_iterations = 10,
  check = FALSE
)
#> Warning: Some expressions had a GC in every iteration; so filtering is
#> disabled.
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 semver      52.96ms  55.78ms      17.5    5.62MB     17.5
#> 2 smvr         8.37ms   9.63ms      97.6  984.45KB     15.9

^{Created on 2025-06-30 with reprex v2.1.1}

[eitsupi]

Slightly faster by #20

version_chrs <- sprintf(
  "1.2.3-foo.%s",
  1:100
)

bench::mark(
  semver = semver::parse_version(version_chrs),
  smvr = smvr::parse_semver(version_chrs),
  min_iterations = 10,
  check = FALSE
)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 semver     149.26µs 185.08µs     4883.    5.53MB     12.0
#> 2 smvr         6.23ms   8.29ms      119.   934.7KB     16.7

^{Created on 2025-07-01 with reprex v2.1.1}