Chasm- Chasm.SemanticVersioning (NuGet), this project, v2.8.1 (Feb 2025);McSherry- McSherry.SemanticVersioning (NuGet) v1.4.1 (Jan 2021);Reeve- SemanticVersioning (NuGet) v3.0.0 (Jan 2025);Hauser- Semver (NuGet) v3.0.0 (Oct 2024);NuGet- NuGet.Versioning (NuGet) v6.12.1 (Nov 2024);
BenchmarkDotNet v0.14.0, Windows 11 (10.0.26100.3037)
AMD Ryzen 5 3500X, 1 CPU, 6 logical and 6 physical cores
.NET SDK 9.0.200-preview.0.25057.12
[Host] : .NET 9.0.0 (9.0.24.52809), X64 RyuJIT AVX2
DefaultJob : .NET 9.0.0 (9.0.24.52809), X64 RyuJIT AVX2
| Method | Categories | Mean | Error | StdDev | Ratio | RatioSD | Gen0 | Allocated | Alloc Ratio |
|---|---|---|---|---|---|---|---|---|---|
| Chasm1 | Sample1 | 208.0 ns | 2.21 ns | 2.07 ns | 1.00 | 0.01 | 0.0381 | 320 B | 1.00 |
| McSherry1 | Sample1 | 1,231.9 ns | 22.82 ns | 46.10 ns | 5.92 | 0.23 | 0.3986 | 3344 B | 10.45 |
| Reeve1 | Sample1 | 2,859.8 ns | 53.82 ns | 47.71 ns | 13.75 | 0.26 | 0.6523 | 5464 B | 17.07 |
| Hauser1 | Sample1 | 1,827.4 ns | 26.52 ns | 24.81 ns | 8.79 | 0.14 | 0.1545 | 1304 B | 4.08 |
| NuGet1 | Sample1 | 766.6 ns | 15.12 ns | 18.00 ns | 3.69 | 0.09 | 0.1650 | 1384 B | 4.33 |
| Chasm2 | Sample2 | 572.9 ns | 11.02 ns | 17.48 ns | 1.00 | 0.04 | 0.1802 | 1512 B | 1.00 |
| McSherry2 | Sample2 | 2,154.9 ns | 26.13 ns | 24.44 ns | 3.76 | 0.12 | 0.5646 | 4736 B | 3.13 |
| Reeve2 | Sample2 | 4,253.1 ns | 46.94 ns | 43.91 ns | 7.43 | 0.23 | 0.9918 | 8312 B | 5.50 |
| Hauser2 | Sample2 | 3,288.7 ns | 31.66 ns | 29.61 ns | 5.75 | 0.18 | 0.3777 | 3184 B | 2.11 |
| NuGet2 | Sample2 | 1,321.5 ns | 25.62 ns | 25.16 ns | 2.31 | 0.08 | 0.3109 | 2616 B | 1.73 |
| Chasm3 | Sample3 | 1,146.4 ns | 18.61 ns | 16.50 ns | 1.00 | 0.02 | 0.3662 | 3064 B | 1.00 |
| McSherry3 | Sample3 | 3,866.3 ns | 71.89 ns | 67.25 ns | 3.37 | 0.07 | 0.8545 | 7168 B | 2.34 |
| Reeve3 | Sample3 | 5,087.8 ns | 100.74 ns | 123.72 ns | 4.44 | 0.12 | 1.1139 | 9360 B | 3.05 |
| Hauser3 | Sample3 | 4,410.9 ns | 86.47 ns | 118.36 ns | 3.85 | 0.11 | 0.6332 | 5304 B | 1.73 |
| NuGet3 | Sample3 | 2,023.3 ns | 40.39 ns | 90.34 ns | 1.77 | 0.08 | 0.4845 | 4072 B | 1.33 |
Chasm significantly outperforms all of the benchmarked alternatives. It's closely followed by NuGet (ratios: 1.77, 2.31, 3.69), then by McSherry (ratios: 3.37, 3.76, 5.92) and Hauser (ratios: 3.85, 5.75, 8.79). Additionally, Chasm allocates as little unnecessary memory as possible during parsing through the use of read-only spans, available on newer framework versions.
| Method | Categories | Mean | Error | StdDev | Ratio | RatioSD | Gen0 | Allocated | Alloc Ratio |
|---|---|---|---|---|---|---|---|---|---|
| Chasm1 | Sample1 | 520.6 ns | 7.33 ns | 6.50 ns | 1.00 | 0.02 | 0.2098 | 1.72 KB | 1.00 |
| McSherry1 | Sample1 | 3,450.2 ns | 32.57 ns | 30.47 ns | 6.63 | 0.10 | 0.8698 | 7.12 KB | 4.14 |
| Reeve1 | Sample1 | 71,406.0 ns | 589.03 ns | 550.98 ns | 137.17 | 2.00 | 9.2773 | 77.32 KB | 44.99 |
| Hauser1 | Sample1 | 1,799.5 ns | 11.57 ns | 10.82 ns | 3.46 | 0.05 | 0.2403 | 1.98 KB | 1.15 |
| Chasm2 | Sample2 | 558.1 ns | 5.80 ns | 5.43 ns | 1.00 | 0.01 | 0.2270 | 1.86 KB | 1.00 |
| McSherry2 | Sample2 | 4,341.9 ns | 10.56 ns | 8.82 ns | 7.78 | 0.08 | 1.1749 | 9.62 KB | 5.17 |
| Reeve2 | Sample2 | 82,737.1 ns | 513.85 ns | 455.51 ns | 148.27 | 1.61 | 11.2305 | 92.38 KB | 49.68 |
| Hauser2 | Sample2 | 2,626.0 ns | 15.04 ns | 11.74 ns | 4.71 | 0.05 | 0.3090 | 2.53 KB | 1.36 |
| Chasm3 | Sample3 | 763.6 ns | 5.02 ns | 4.69 ns | 1.00 | 0.01 | 0.3109 | 2.55 KB | 1.00 |
| McSherry3 | Sample3 | 6,415.8 ns | 15.51 ns | 14.51 ns | 8.40 | 0.05 | 1.7471 | 14.29 KB | 5.61 |
| Reeve3 | Sample3 | 150,100.6 ns | 1,006.82 ns | 941.78 ns | 196.57 | 1.68 | 20.0195 | 164.57 KB | 64.62 |
| Hauser3 | Sample3 | 4,466.0 ns | 22.92 ns | 21.44 ns | 5.85 | 0.04 | 0.4501 | 3.7 KB | 1.45 |
| Chasm4 | Sample4 | 2,096.0 ns | 19.48 ns | 17.27 ns | 1.00 | 0.01 | 0.7629 | 6.26 KB | 1.00 |
| McSherry4 | Sample4 | 16,489.3 ns | 148.90 ns | 139.28 ns | 7.87 | 0.09 | 4.4556 | 36.6 KB | 5.85 |
| Reeve4 | Sample4 | 351,562.4 ns | 2,418.91 ns | 2,019.90 ns | 167.74 | 1.64 | 44.9219 | 377.77 KB | 60.37 |
| Hauser4 | Sample4 | 12,782.5 ns | 80.67 ns | 75.46 ns | 6.10 | 0.06 | 1.1902 | 9.82 KB | 1.57 |
Firstly, some libraries don't fully support node-semver's syntax. McSherry doesn't recognize wildcards (e.g. =1.2.x, 1.2 - 3.4), and both Hauser and Reeve don't parse hyphen ranges in complex sets (e.g. >1.2.0 1.5 - 1.7). Chasm though, as far as I'm aware, follows node-semver's specifications exactly.
Chasm outperforms other libraries once again, with Hauser (ratios: 3.46, 4.71, 5.85, 6.10) and McSherry (ratios: 6.63, 7.78, 8.40, 7.87) pretty far behind. Reeve's performance is particularly bad, with ratios: 137, 148, 196, 167. Both Chasm and Hauser make little allocations, with Hauser allocating slightly more.
| Method | Categories | Mean | Error | StdDev | Ratio | RatioSD | Gen0 | Allocated | Alloc Ratio |
|---|---|---|---|---|---|---|---|---|---|
| Chasm1 | Sample1 | 120.782 ns | 1.0064 ns | 0.9413 ns | 1.00 | 0.01 | 0.0248 | 208 B | 1.00 |
| McSherry1 | Sample1 | 385.063 ns | 4.6027 ns | 4.3054 ns | 3.19 | 0.04 | 0.1469 | 1232 B | 5.92 |
| Reeve1 | Sample1 | 7.424 ns | 0.0206 ns | 0.0193 ns | 0.06 | 0.00 | - | - | 0.00 |
| Hauser1 | Sample1 | 518.714 ns | 3.0363 ns | 2.8401 ns | 4.29 | 0.04 | 0.0954 | 800 B | 3.85 |
| NuGet1 | Sample1 | 278.223 ns | 0.9988 ns | 0.8854 ns | 2.30 | 0.02 | 0.0248 | 208 B | 1.00 |
| Chasm2 | Sample2 | 176.107 ns | 3.5648 ns | 6.1491 ns | 1.00 | 0.05 | 0.0439 | 368 B | 1.00 |
| McSherry2 | Sample2 | 769.424 ns | 15.1324 ns | 24.0017 ns | 4.37 | 0.20 | 0.2193 | 1840 B | 5.00 |
| Reeve2 | Sample2 | 7.555 ns | 0.0440 ns | 0.0412 ns | 0.04 | 0.00 | - | - | 0.00 |
| Hauser2 | Sample2 | 551.615 ns | 6.2294 ns | 5.8269 ns | 3.14 | 0.11 | 0.1297 | 1088 B | 2.96 |
| NuGet2 | Sample2 | 424.241 ns | 8.2521 ns | 7.7190 ns | 2.41 | 0.09 | 0.0715 | 600 B | 1.63 |
| Chasm3 | Sample3 | 300.364 ns | 4.6458 ns | 4.3456 ns | 1.00 | 0.02 | 0.0715 | 600 B | 1.00 |
| McSherry3 | Sample3 | 1,252.509 ns | 5.8264 ns | 5.1649 ns | 4.17 | 0.06 | 0.3281 | 2744 B | 4.57 |
| Reeve3 | Sample3 | 7.534 ns | 0.0448 ns | 0.0419 ns | 0.03 | 0.00 | - | - | 0.00 |
| Hauser3 | Sample3 | 583.903 ns | 6.7078 ns | 5.9463 ns | 1.94 | 0.03 | 0.1822 | 1528 B | 2.55 |
| NuGet3 | Sample3 | 446.683 ns | 8.5906 ns | 8.0356 ns | 1.49 | 0.03 | 0.0772 | 648 B | 1.08 |
Reeve stores the original string, resulting in a simple field read. Then, Chasm is in second, followed by NuGet (ratios: 1.49, 2.41, 2.30) and Hauser (ratios: 1.94, 3.14, 4.29). McSherry is far behind. Additionally, Chasm allocates as little unnecessary memory as possible; NuGet behaves similarly, but it's not ideal.
| Library | .NET Core | .NET Standard | .NET Framework |
|---|---|---|---|
| McSherry | .NET Core 1.0 | .NET Standard 1.0 | .NET Framework 4.5 |
| Chasm | .NET Core 2.0 | .NET Standard 2.0 | .NET Framework 4.6.1 |
| Reeve | .NET Core 2.0 | .NET Standard 2.0 | .NET Framework 4.6.1 |
| Hauser | .NET Core 2.0 | .NET Standard 2.0 | .NET Framework 4.6.1 |
| NuGet | .NET Core 2.0 | .NET Standard 2.0 | .NET Framework 4.6.1 |
- explicitly included target framework (in package)
- compatible target framework (through .NET Standard)
| Library | Raw total | Packaged |
|---|---|---|
| Chasm | 96.0 KiB | 37.7 KiB |
| McSherry | 50.0 KiB | 19.8 KiB |
| Reeve | 34.0 KiB | 13.8 KiB |
| Hauser | 71.0 KiB | 27.3 KiB |
| NuGet | 62.6 KiB | 31.3 KiB |
The size differences are negligible, they're all roughly the size of a low quality meme PNG.