Skip to content

bench: update @tuananh/sax-parser to 1.4 and police arity probing #17

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
evoactivity merged 1 commit into from
Jul 19, 2026
Merged

bench: update @tuananh/sax-parser to 1.4 and police arity probing #17

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
144 changes: 72 additions & 72 deletions eksml/BENCHMARKS.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,66 +105,66 @@ Chunked streaming parse where each parser tokenizes SAX events and builds a full
</tr>
<tr>
<td><strong>Eksml (SAX)</strong></td>
<td align="right"><strong>95,408 op/s</strong></td>
<td align="right"><strong>17,945 op/s</strong></td>
<td align="right"><strong>13,932 op/s</strong></td>
<td align="right"><strong>14,932 op/s</strong></td>
<td align="right"><strong>13,431 op/s</strong></td>
<td align="right"><strong>93,671 op/s</strong></td>
<td align="right"><strong>17,666 op/s</strong></td>
<td align="right"><strong>14,402 op/s</strong></td>
<td align="right"><strong>14,708 op/s</strong></td>
<td align="right"><strong>13,372 op/s</strong></td>
</tr>
<tr>
<td>easysax</td>
<td align="right">66,897 op/s</td>
<td align="right">14,591 op/s</td>
<td align="right">11,235 op/s</td>
<td align="right">10,517 op/s</td>
<td align="right">10,604 op/s</td>
<td align="right">61,700 op/s</td>
<td align="right">14,253 op/s</td>
<td align="right">11,063 op/s</td>
<td align="right">10,639 op/s</td>
<td align="right">10,204 op/s</td>
</tr>
<tr>
<td>Eksml (XmlParseStream)</td>
<td align="right">30,549 op/s</td>
<td align="right">9,794 op/s</td>
<td align="right">8,947 op/s</td>
<td align="right">4,457 op/s</td>
<td align="right">8,690 op/s</td>
<td align="right">28,981 op/s</td>
<td align="right">9,531 op/s</td>
<td align="right">8,924 op/s</td>
<td align="right">4,670 op/s</td>
<td align="right">8,646 op/s</td>
</tr>
<tr>
<td>htmlparser2</td>
<td align="right">30,083 op/s</td>
<td align="right">7,384 op/s</td>
<td align="right">5,461 op/s</td>
<td align="right">6,912 op/s</td>
<td align="right">6,877 op/s</td>
<td align="right">28,402 op/s</td>
<td align="right">6,277 op/s</td>
<td align="right">5,386 op/s</td>
<td align="right">6,807 op/s</td>
<td align="right">6,802 op/s</td>
</tr>
<tr>
<td>saxes</td>
<td align="right">26,185 op/s</td>
<td align="right">6,602 op/s</td>
<td align="right">6,780 op/s</td>
<td align="right">5,821 op/s</td>
<td align="right">5,634 op/s</td>
<td align="right">27,406 op/s</td>
<td align="right">6,519 op/s</td>
<td align="right">6,732 op/s</td>
<td align="right">5,510 op/s</td>
<td align="right">4,499 op/s</td>
</tr>
<tr>
<td>sax</td>
<td align="right">14,988 op/s</td>
<td align="right">3,561 op/s</td>
<td align="right">3,044 op/s</td>
<td align="right">3,806 op/s</td>
<td align="right">2,922 op/s</td>
<td>@tuananh/sax-parser</td>
<td align="right">25,737 op/s</td>
<td align="right">6,579 op/s</td>
<td align="right">6,085 op/s</td>
<td align="right">6,196 op/s</td>
<td align="right">4,189 op/s</td>
</tr>
<tr>
<td>@tuananh/sax-parser</td>
<td align="right">14,084 op/s</td>
<td align="right">3,918 op/s</td>
<td align="right">3,249 op/s</td>
<td align="right">2,394 op/s</td>
<td align="right">2,410 op/s</td>
<td>sax</td>
<td align="right">14,841 op/s</td>
<td align="right">3,550 op/s</td>
<td align="right">3,016 op/s</td>
<td align="right">3,792 op/s</td>
<td align="right">2,909 op/s</td>
</tr>
</table>

Eksml's SAX parser is **1.2-1.4x faster than easysax**, **2-3.6x faster than htmlparser2/saxes**, and **3.9-6.4x faster than sax**.
Eksml's SAX parser is **1.2-1.5x faster than easysax**, **2-3.4x faster than htmlparser2/saxes**, and **3.9-6.3x faster than sax**.

> [!note]
> @tuananh/sax-parser is a native C++ addon; every `write()` crosses the JS↔C++ boundary, which dominates in chunked streaming.
> @tuananh/sax-parser is a native C++ addon; marshalling event arguments across the JS↔C++ boundary dominates its chunked-streaming cost. Since 1.4 it skips argument materialization for zero-arity listeners, so the no-op benchmark callbacks declare parameters — every other parser materializes arguments unconditionally, and the comparison requires equal work.

## Raw Tokenization (no-op callbacks)

Expand All @@ -181,55 +181,55 @@ Pure scanner throughput with no downstream work -- isolates the tokenizer's raw
</tr>
<tr>
<td><strong>Eksml (SAX)</strong></td>
<td align="right"><strong>104,122 op/s</strong></td>
<td align="right"><strong>20,970 op/s</strong></td>
<td align="right"><strong>19,927 op/s</strong></td>
<td align="right"><strong>17,063 op/s</strong></td>
<td align="right"><strong>15,605 op/s</strong></td>
<td align="right"><strong>113,482 op/s</strong></td>
<td align="right"><strong>20,462 op/s</strong></td>
<td align="right"><strong>18,506 op/s</strong></td>
<td align="right"><strong>17,186 op/s</strong></td>
<td align="right"><strong>17,462 op/s</strong></td>
</tr>
<tr>
<td>easysax</td>
<td align="right">73,389 op/s</td>
<td align="right">15,678 op/s</td>
<td align="right">11,899 op/s</td>
<td align="right">11,982 op/s</td>
<td align="right">11,933 op/s</td>
<td align="right">74,171 op/s</td>
<td align="right">15,868 op/s</td>
<td align="right">12,637 op/s</td>
<td align="right">11,124 op/s</td>
<td align="right">11,316 op/s</td>
</tr>
<tr>
<td>saxes</td>
<td align="right">35,355 op/s</td>
<td align="right">9,492 op/s</td>
<td align="right">7,290 op/s</td>
<td align="right">8,870 op/s</td>
<td align="right">8,562 op/s</td>
<td align="right">33,493 op/s</td>
<td align="right">9,434 op/s</td>
<td align="right">8,241 op/s</td>
<td align="right">7,690 op/s</td>
<td align="right">7,885 op/s</td>
</tr>
<tr>
<td>htmlparser2</td>
<td align="right">30,520 op/s</td>
<td align="right">7,647 op/s</td>
<td align="right">6,334 op/s</td>
<td align="right">6,475 op/s</td>
<td align="right">7,145 op/s</td>
<td align="right">28,743 op/s</td>
<td align="right">7,347 op/s</td>
<td align="right">6,309 op/s</td>
<td align="right">7,122 op/s</td>
<td align="right">7,081 op/s</td>
</tr>
<tr>
<td>sax</td>
<td align="right">15,307 op/s</td>
<td align="right">3,679 op/s</td>
<td align="right">3,184 op/s</td>
<td align="right">3,928 op/s</td>
<td align="right">3,315 op/s</td>
<td>@tuananh/sax-parser</td>
<td align="right">27,172 op/s</td>
<td align="right">7,496 op/s</td>
<td align="right">6,145 op/s</td>
<td align="right">6,001 op/s</td>
<td align="right">5,036 op/s</td>
</tr>
<tr>
<td>@tuananh/sax-parser</td>
<td align="right">12,907 op/s</td>
<td align="right">4,197 op/s</td>
<td align="right">3,217 op/s</td>
<td align="right">2,388 op/s</td>
<td align="right">2,608 op/s</td>
<td>sax</td>
<td align="right">15,458 op/s</td>
<td align="right">3,997 op/s</td>
<td align="right">2,945 op/s</td>
<td align="right">3,664 op/s</td>
<td align="right">3,252 op/s</td>
</tr>
</table>

Eksml's tokenizer is **1.3-1.7x faster than easysax**, **1.8-3.4x faster than saxes/htmlparser2**, and **4.3-6.8x faster than sax**.
Eksml's tokenizer is **1.3-1.6x faster than easysax**, **2.2-4x faster than saxes/htmlparser2**, and **4.7-7.3x faster than sax**.

> [!note]
> easysax parses attributes lazily (`startNode` receives a `getAttr()` thunk); the benchmark invokes it so every parser materializes attributes, the work all other parsers do unconditionally for their open-tag events.
Expand Down
4 changes: 2 additions & 2 deletions eksml/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -534,8 +534,8 @@ import {
Eksml is consistently the fastest at parsing, streaming, and tokenization, and leads serialization with `validate: false`. Benchmarks run via [Vitest bench](https://vitest.dev/guide/features.html#benchmarking) against real-world XML fixtures from ~100 B to ~30 KB, plus a synthetic attribute-heavy stress document.

- **DOM parsing**: 1.7-2.2x faster than tXml, 3-6x faster than htmlparser2, 10-25x faster than fast-xml-parser/xml2js/xmldom
- **SAX streaming**: 1.2-1.4x faster than easysax, 2-3.6x faster than htmlparser2/saxes, 3.9-6.4x faster than sax
- **Raw tokenization**: 1.3-1.7x faster than easysax, 1.8-3.4x faster than saxes/htmlparser2, 4.3-6.8x faster than sax
- **SAX streaming**: 1.2-1.5x faster than easysax, 2-3.4x faster than htmlparser2/saxes, 3.9-6.3x faster than sax
- **Raw tokenization**: 1.3-1.6x faster than easysax, 2.2-4x faster than saxes/htmlparser2, 4.7-7.3x faster than sax
- **Serialization**: Fastest on every fixture with `validate: false`; with validation on (the default), wins RSS/SOAP/Atom/POM while tXml edges small/EPG; 2-7x faster than xmldom and 7-19x faster than fast-xml-parser/xml2js

Full results with per-fixture op/s tables: **[BENCHMARKS.md](https://github.com/evoactivity/eksml/blob/main/eksml/BENCHMARKS.md)**
Expand Down
3 changes: 2 additions & 1 deletion eksml/bench/stream.bench.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -238,7 +238,8 @@ tuananhParser.on('text', (text: string) => {
tuananhParser.on('cdata', (cdata: string) => {
pushChild(cdata);
});
tuananhParser.on('endElement', () => {
// Parameter declared so arity probing cannot skip materializing the name
tuananhParser.on('endElement', (_name: string) => {
stack.pop();
});

Expand Down
7 changes: 6 additions & 1 deletion eksml/bench/tokenize.bench.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,7 +66,12 @@ const xmltvChunks64 = chunkString(xmltvEpg, 64);
const attrHeavyChunks256 = chunkString(attrHeavy, 256);

// No-op function used as callback
const noop = () => {};
// No-op callback with DECLARED parameters. Some parsers probe listener arity
// (Function.length) and skip materializing event arguments for zero-arity
// handlers (@tuananh/sax-parser 1.4 inflates ~3x this way). Every JS parser
// here materializes arguments unconditionally, so the no-op declares
// parameters to keep the measured work equal.
const noop = (_a?: unknown, _b?: unknown) => {};

// ---------------------------------------------------------------------------
// eksml createSaxParser — persistent parser, no-op callbacks (close() resets)
Expand Down
2 changes: 1 addition & 1 deletion eksml/package.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,7 +82,7 @@
"license": "MIT",
"packageManager": "pnpm@10.33.0",
"devDependencies": {
"@tuananh/sax-parser": "^1.2.0",
"@tuananh/sax-parser": "^1.4.0",
"@types/node": "^25.5.0",
"@types/sax": "^1.2.7",
"@types/xml2js": "^0.4.14",
Expand Down
Loading