Skip to content

Commit be3c7ef

Browse files
vinitkumarvinitkumar
committed
chore: release v6.0.1
- Enable 'fast' optional dependency: pip install json2xml[fast] - Add comprehensive benchmark comparing Python/Rust/Go/Zig - Update BENCHMARKS.md with latest results showing: - Rust: ~28x faster than Python - Go: 48x faster for large files (300KB+) - Zig: 3-6x faster for medium-large files
1 parent ccb72d3 commit be3c7ef

File tree

5 files changed

+474
-80
lines changed

5 files changed

+474
-80
lines changed

‎BENCHMARKS.md‎

Lines changed: 85 additions & 79 deletions
Original file line numberDiff line numberDiff line change
@@ -1,127 +1,133 @@
11
# json2xml Benchmark Results
22

3-
Comprehensive performance comparison between Python implementations and the Go version of json2xml.
3+
Comprehensive performance comparison between all json2xml implementations.
44

55
## Test Environment
66

7-
- **Machine**: Apple Silicon (aarch64)
7+
- **Machine**: Apple Silicon (M-series, aarch64)
88
- **OS**: macOS
9-
- **Date**: January 14, 2026
9+
- **Date**: January 16, 2026
1010

1111
### Implementations Tested
1212

13-
| Implementation | Version | Notes |
14-
|----------------|---------|-------|
15-
| CPython | 3.14.2 | Homebrew installation |
16-
| CPython | 3.15.0a4 | Latest alpha via uv |
17-
| PyPy | 3.10.16 | JIT-compiled Python |
18-
| Go | 1.0.0 | json2xml-go |
13+
| Implementation | Type | Notes |
14+
|----------------|------|-------|
15+
| Python | Library | Pure Python (json2xml) |
16+
| Rust | Library | Native extension via PyO3 (json2xml-rs) |
17+
| Go | CLI | Standalone binary (json2xml-go) |
18+
| Zig | CLI | Standalone binary (json2xml-zig) |
1919

2020
## Test Data
2121

2222
| Size | Description | Bytes |
2323
|------|-------------|-------|
2424
| Small | Simple object `{"name": "John", "age": 30, "city": "New York"}` | 47 |
25-
| Medium | `bigexample.json` (patent data) | 2,598 |
26-
| Large | 1,000 generated records with nested structures | 323,130 |
27-
| Very Large | 5,000 generated records with nested structures | 1,619,991 |
25+
| Medium | 10 generated records with nested structures | 3,212 |
26+
| bigexample.json | Real-world patent data | 2,018 |
27+
| Large | 100 generated records with nested structures | 32,226 |
28+
| Very Large | 1,000 generated records with nested structures | 323,126 |
2829

2930
## Results
3031

31-
### Individual Test Results
32+
### Performance Summary
3233

33-
| Test | CPython 3.14.2 | CPython 3.15.0a4 | PyPy 3.10.16 | Go |
34-
|------|----------------|------------------|--------------|-----|
35-
| **Small JSON** (47 bytes) | 75.46ms | 55.74ms (**1.4x faster**) | 121.47ms (1.6x slower) | 3.69ms (**20.4x faster**) |
36-
| **Medium JSON** (2.6KB) | 73.87ms | 57.98ms (**1.3x faster**) | 125.73ms (1.7x slower) | 4.32ms (**17.1x faster**) |
37-
| **Large JSON** (323KB) | 419.67ms | 328.98ms (**1.3x faster**) | 517.51ms (1.2x slower) | 67.13ms (**6.3x faster**) |
38-
| **Very Large JSON** (1.6MB) | 2.09s | 1.86s (**1.1x faster**) | 1.42s (**1.5x faster**) | 287.58ms (**7.3x faster**) |
34+
| Test Case | Python | Rust | Go | Zig |
35+
|-----------|--------|------|-----|-----|
36+
| Small (47B) | 40.12µs | 1.45µs | 4.65ms | 3.74ms |
37+
| Medium (3.2KB) | 2.14ms | 71.28µs | 4.07ms | 3.28ms |
38+
| bigexample (2KB) | 819.46µs | 32.88µs | 4.02ms | 2.96ms |
39+
| Large (32KB) | 21.08ms | 739.89µs | 4.05ms | 6.11ms |
40+
| Very Large (323KB) | 212.61ms | 7.55ms | 4.38ms | 33.24ms |
3941

40-
### Summary (Average Across All Tests)
42+
### Speedup vs Pure Python
4143

42-
| Implementation | Avg Time | vs CPython 3.14.2 |
43-
|----------------|----------|-------------------|
44-
| **Go** | 90.68ms | **7.34x faster** 🚀 |
45-
| **PyPy 3.10.16** | 545.58ms | **1.22x faster** |
46-
| **CPython 3.15.0a4** | 575.45ms | **1.16x faster** |
47-
| **CPython 3.14.2** | 665.23ms | baseline |
44+
| Test Case | Rust | Go | Zig |
45+
|-----------|------|-----|-----|
46+
| Small (47B) | **27.6x** | 0.0x* | 0.0x* |
47+
| Medium (3.2KB) | **30.0x** | 0.5x* | 0.7x* |
48+
| bigexample (2KB) | **24.9x** | 0.2x* | 0.3x* |
49+
| Large (32KB) | **28.5x** | 5.2x | 3.5x |
50+
| Very Large (323KB) | **28.2x** | **48.5x** | 6.4x |
51+
52+
*CLI tools have process spawn overhead (~3-4ms) which dominates for small inputs
4853

4954
## Key Observations
5055

51-
### 1. Go is the Clear Winner
56+
### 1. Rust Extension is the Best Choice for Python Users 🦀
5257

53-
Go outperforms all Python implementations by a significant margin:
54-
- **7.34x faster** than CPython 3.14.2 on average
55-
- Up to **20x faster** for small inputs due to minimal startup overhead
56-
- Consistent performance across all input sizes
58+
The Rust extension (json2xml-rs) provides:
59+
- **~28x faster** than pure Python consistently across all input sizes
60+
- **Zero process overhead** - called directly from Python
61+
- **Automatic fallback** - pure Python used if Rust unavailable
62+
- **Easy install**: `pip install json2xml[fast]`
5763

58-
### 2. CPython 3.15.0a4 Shows Promising Improvements
64+
### 2. Go Excels for Large CLI Workloads 🚀
5965

60-
The latest Python alpha demonstrates consistent performance gains:
61-
- **13-35% faster** than CPython 3.14.2 across all test sizes
62-
- Improvements likely due to ongoing interpreter optimizations
66+
For very large inputs (323KB+):
67+
- **48.5x faster** than Python
68+
- But ~3-4ms startup overhead hurts small file performance
69+
- Best for batch processing or large file conversions
6370

64-
### 3. PyPy Has Interesting Trade-offs
71+
### 3. Zig is Competitive but Has Trade-offs
6572

66-
PyPy's JIT compiler creates a unique performance profile:
67-
- **Slower for small/medium inputs**: JIT compilation overhead hurts for quick operations
68-
- **Faster for very large inputs**: JIT shines on the 5K record test (1.5x faster than CPython)
69-
- Best suited for long-running processes or batch processing
73+
- Consistent ~3ms startup overhead
74+
- Good for medium-large files (3-6x faster than Python)
75+
- Less optimized than Go for very large inputs
7076

71-
### 4. Startup Overhead Dominates Small Inputs
77+
### 4. Process Spawn Overhead Matters
7278

73-
Python's interpreter startup time is significant:
74-
- CPython takes **55-75ms** even for 47 bytes of JSON
75-
- Go takes only **3.7ms** for the same operation
76-
- For CLI tools processing small files, Go provides a much better user experience
79+
CLI tools (Go, Zig) have ~3-4ms process spawn overhead:
80+
- Dominates for small inputs (makes them appear slower than Python!)
81+
- Negligible for large inputs where actual work dominates
82+
- Rust extension avoids this entirely by being a native Python module
7783

7884
## When to Use Each Implementation
7985

80-
| Use Case | Recommended |
81-
|----------|-------------|
82-
| CLI tool for small/medium files | **Go** (json2xml-go) |
83-
| High-throughput batch processing | **Go** or **PyPy** |
84-
| Integration with Python codebase | **CPython 3.15+** |
85-
| One-off conversions in scripts | **CPython** (any version) |
86-
87-
## Running the Benchmarks
86+
| Use Case | Recommended | Why |
87+
|----------|-------------|-----|
88+
| Python library calls | **Rust** (`pip install json2xml[fast]`) | 28x faster, no overhead |
89+
| Small files via CLI | **Rust** via Python | CLI overhead dominates |
90+
| Large files via CLI | **Go** (json2xml-go) | 48x faster for 300KB+ |
91+
| Batch processing | **Go** or **Rust** | Both excellent |
92+
| Pure Python required | **Python** (json2xml) | Always available |
8893

89-
### Python Multi-Implementation Benchmark
94+
## Installation
9095

9196
```bash
92-
# Set the Go CLI path
93-
export JSON2XML_GO_CLI=/path/to/json2xml-go
97+
# Pure Python (always works)
98+
pip install json2xml
9499

95-
# Run the benchmark
96-
python benchmark_multi_python.py
100+
# With Rust acceleration (recommended)
101+
pip install json2xml[fast]
102+
103+
# Go CLI
104+
go install github.com/vinitkumar/json2xml-go@latest
105+
106+
# Zig CLI
107+
# See: github.com/nicholasgriffintn/json2xml-zig
97108
```
98109

99-
### Simple Python vs Go Benchmark
110+
## Running the Benchmarks
111+
112+
### Comprehensive Benchmark (All Implementations)
100113

101114
```bash
102-
# Set paths via environment variables (optional)
103-
export JSON2XML_GO_CLI=/path/to/json2xml-go
104-
export JSON2XML_EXAMPLES_DIR=/path/to/examples
115+
python benchmark_all.py
116+
```
105117

106-
# Run the benchmark
107-
python benchmark.py
118+
### Rust vs Python Only
119+
120+
```bash
121+
python benchmark_rust.py
108122
```
109123

110-
## Reproducing Results
124+
### Multi-Python Version Benchmark
111125

112-
1. Install required Python versions using `uv`:
113-
```bash
114-
uv python install 3.14 3.15.0a4 [email protected]
115-
```
126+
```bash
127+
python benchmark_multi_python.py
128+
```
116129

117-
2. Build the Go binary:
118-
```bash
119-
cd /path/to/json2xml-go
120-
go build -o json2xml-go ./cmd/json2xml-go
121-
```
130+
## Related Projects
122131

123-
3. Run the multi-Python benchmark:
124-
```bash
125-
cd /path/to/json2xml
126-
python benchmark_multi_python.py
127-
```
132+
- **Go version**: [github.com/vinitkumar/json2xml-go](https://github.com/vinitkumar/json2xml-go)
133+
- **Zig version**: [github.com/nicholasgriffintn/json2xml-zig](https://github.com/nicholasgriffintn/json2xml-zig)

‎HISTORY.rst‎

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,18 @@
11
History
22
=======
33

4+
6.0.1 / 2026-01-16
5+
==================
6+
7+
* feat: enable 'fast' optional dependency for easy Rust installation
8+
* docs: update benchmarks with comprehensive Python/Rust/Go/Zig comparison
9+
* feat: add benchmark_all.py for multi-implementation benchmarking
10+
11+
Installation with Rust acceleration:
12+
13+
* pip install json2xml[fast]
14+
15+
416
6.0.0 / 2026-01-16
517
==================
618

0 commit comments

Comments
 (0)