docs: Fix benchmark chart naming and add data migration script #62
Conversation
The benchmark chart names were changed in commit 7bd08a2 from: - 'Virtual DOM Benchmarks' → 'Rendering Engine Throughput' - 'Virtual DOM Allocations' → 'Rendering Engine Allocations' This caused the github-action-benchmark to create new chart sets while the old ones stopped receiving updates. The fix has been applied to gh-pages (commit b3b8465) to merge historical data. Changes: - Update docs/benchmarks.md to reflect new naming (3 benchmark categories) - Add scripts/fix-benchmark-data.py for merging chart set data - Document the new benchmark categories (Diffing, Rendering, Handlers)
There was a problem hiding this comment.
Pull request overview
Updates benchmark documentation to reflect the new GitHub Pages chart naming and introduces a small migration utility to merge historical github-action-benchmark chart data when chart names change.
Changes:
- Add
scripts/fix-benchmark-data.pyto merge legacy chart sets into the new chart names. - Update
docs/benchmarks.mdto rename the benchmarks doc and describe benchmark categories. - Refresh the benchmarks “Architecture” section to list the benchmark entry points.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| scripts/fix-benchmark-data.py | New Python utility to merge old github-action-benchmark entry sets into renamed chart sets. |
| docs/benchmarks.md | Updates benchmark documentation naming and structure, and adds category overview. |
| # Read the data.js file | ||
| content = Path(input_file).read_text() | ||
|
|
||
| # Remove the JavaScript wrapper | ||
| if content.startswith("window.BENCHMARK_DATA = "): | ||
| json_str = content.replace("window.BENCHMARK_DATA = ", "", 1) | ||
| else: | ||
| json_str = content | ||
|
|
||
| data = json.loads(json_str) |
There was a problem hiding this comment.
Path(input_file).read_text() uses the platform default encoding and can fail on machines where data.js contains non-ASCII characters (e.g., the ± symbols you mention). Read with an explicit UTF-8 encoding, and consider stripping an optional trailing semicolon/newline after removing the window.BENCHMARK_DATA = prefix so json.loads() works with the common ... = {...}; form too.
|
|
||
| # Sort each entry list by date | ||
| for name in entries: | ||
| entries[name].sort(key=lambda e: e.get("date", 0)) |
There was a problem hiding this comment.
The sort key uses e.get("date", 0). If any entry is missing date, this returns an int while other entries likely have a string/ISO date, and Python will raise a TypeError when sorting mixed key types. Use a consistent key type (e.g., empty string default, or parse dates to a numeric timestamp) to make the script robust to missing/invalid date values.
| entries[name].sort(key=lambda e: e.get("date", 0)) | |
| entries[name].sort(key=lambda e: e.get("date", "")) |
| ### DOM Diffing (`Abies.Benchmarks.Diffing/`) | ||
|
|
||
| Measures the Virtual DOM diffing algorithm performance. | ||
|
|
||
| ### Rendering (`Abies.Benchmarks.Rendering/`) | ||
|
|
||
| Measures HTML string rendering performance. | ||
|
|
||
| ### Event Handlers (`Abies.Benchmarks.Handlers/`) |
There was a problem hiding this comment.
The category headings reference directories like Abies.Benchmarks.Diffing/, Abies.Benchmarks.Rendering/, and Abies.Benchmarks.Handlers/, but the repository’s benchmarks appear to be a single Abies.Benchmarks/ project with DomDiffingBenchmarks.cs, RenderingBenchmarks.cs, and EventHandlerBenchmarks.cs (no such subfolders). Update these references to match the actual structure (e.g., point at the benchmark classes/files, or the workflow artifact folders benchmark-results/diffing|rendering|handlers).
| ### DOM Diffing (`Abies.Benchmarks.Diffing/`) | |
| Measures the Virtual DOM diffing algorithm performance. | |
| ### Rendering (`Abies.Benchmarks.Rendering/`) | |
| Measures HTML string rendering performance. | |
| ### Event Handlers (`Abies.Benchmarks.Handlers/`) | |
| ### DOM Diffing (`Abies.Benchmarks/DomDiffingBenchmarks.cs`) | |
| Measures the Virtual DOM diffing algorithm performance. | |
| ### Rendering (`Abies.Benchmarks/RenderingBenchmarks.cs`) | |
| Measures HTML string rendering performance. | |
| ### Event Handlers (`Abies.Benchmarks/EventHandlerBenchmarks.cs`) |
1 participant
📝 Description
What
Fixes the benchmark chart naming mismatch on GitHub Pages and adds a utility script for future data migrations.
Why
The benchmark chart names were changed in commit
7bd08a2from:This caused
github-action-benchmarkto create new chart sets while the old ones stopped receiving updates. The interactive charts at https://picea.github.io/Abies/dev/bench/ showed 4 separate charts (2 stale + 2 current) instead of 2 unified charts with complete history.How
b3b8465): Merged historical data from old chart sets into new ones using the fix scriptdocs/benchmarks.mdto reflect the new naming and benchmark categoriesscripts/fix-benchmark-data.pyfor future migrations🔗 Related Issues
N/A - discovered during routine monitoring
✅ Type of Change
🧪 Testing
Test Coverage
Testing Details
https://raw.githubusercontent.com/Picea/Abies/gh-pages/dev/bench/data.jsand confirmed merge was successful✨ Changes Made
docs/benchmarks.md: Updated title from "Virtual DOM Benchmarks" to "Rendering Engine Benchmarks", added documentation for 3 benchmark categories (Diffing, Rendering, Handlers), updated Architecture sectionscripts/fix-benchmark-data.py: New utility script to merge benchmark chart data between name changes🔍 Code Review Checklist