by axew3 » Sun Aug 30, 2026 3:13 pm
This is actually the 3.1.0 update Changelog.
It will be presumably released today.
Code: Select all
== Changelog ==
= 3.1.0 =
*Release Date - 30 Aug, 2026*
* Update: Upgraded the Noble post-quantum cryptographic library to the latest stable version 0.7.1.
* Compatibility: Old existing ciphered files remain (obviously) fully backward compatible with this update.
* Technical details: https://www.axew3.com/w3/forums/viewtopic.php?t=2189
* Remove: Removed the public function `remove_html_and_js_comments($content)`. This function was previously executed to prevent WordPress `wpautop()` from automatically injecting `<p>` tags into the embedded `w3mypq_body.html` file. This issue is now bypassed by properly wrapping all HTML tags and HTML comments inside container `<div>` elements instead into the `w3mypq_body.html` file.
Noble 0.7.1 (run better):
The technical changes that directly cause this smoother performance include:
1. Synchronous Refactoring (No Event Loop Overhead)
In earlier experimental versions, many cryptographic steps utilized asynchronous promises (async/await) or internal micro-task queuing. In version 0.7.1, core primitives like encapsulate() and decapsulate() execute completely synchronously. By avoiding the JavaScript Event Loop overhead, the browser/worker doesn't pause to schedule promises, resulting in immediate execution loops.
2. Zero-Copy Garbage Collection (Memory Fluidity)
Post-quantum cryptography is mathematically heavy and handles enormous matrices of bytes.
Version 0.7.1 optimized how Uint8Array buffers are recycled internally:
Fewer Array Allocations: The library now reuses memory arrays rather than constantly instantiating new intermediate arrays in RAM.Reduced GC Pauses: Because fewer temporary arrays are created, the browser's Garbage Collector (GC) doesn't have to pause execution to clean up memory junk, preventing micro-stutters during heavy encryption loops.
3. Loop Unrolling and Optimized NTTRef (Number Theoretic Transform)
Both ML-KEM and ML-DSA rely on polynomial multiplications handled via the Number Theoretic Transform (NTT). The math engine in 0.7.1 was tightly optimized:
Loop Unrolling: Critical mathematical loops were unrolled to execute consecutive assembly-like operations rather than running thousands of generic iterations.
Bitwise Efficiency: Core operations leverage native JavaScript 32-bit integer boundaries more efficiently, which modern V8 engines (Chrome, Edge, Node.js) optimize flawlessly.
4. Leaner Tree-ShakingBy modularizing the library paths (@noble/post-quantum/ml-kem.js), the bundle compiled by esbuild strips away all unneeded cryptographic overhead (like Falcon or Sphincs+ parameters). This ensures the CPU cache only deals with the lean, targeted binary instructions of your two specific modules.
This is actually the 3.1.0 update Changelog.
It will be presumably released today.
[code]
== Changelog ==
= 3.1.0 =
*Release Date - 30 Aug, 2026*
* Update: Upgraded the Noble post-quantum cryptographic library to the latest stable version 0.7.1.
* Compatibility: Old existing ciphered files remain (obviously) fully backward compatible with this update.
* Technical details: https://www.axew3.com/w3/forums/viewtopic.php?t=2189
* Remove: Removed the public function `remove_html_and_js_comments($content)`. This function was previously executed to prevent WordPress `wpautop()` from automatically injecting `<p>` tags into the embedded `w3mypq_body.html` file. This issue is now bypassed by properly wrapping all HTML tags and HTML comments inside container `<div>` elements instead into the `w3mypq_body.html` file.[/code]
Noble 0.7.1 (run better):
[b]The technical changes that directly cause this smoother performance include:[/b]
1. Synchronous Refactoring (No Event Loop Overhead)
In earlier experimental versions, many cryptographic steps utilized asynchronous promises (async/await) or internal micro-task queuing. In version 0.7.1, core primitives like encapsulate() and decapsulate() execute completely synchronously. By avoiding the JavaScript Event Loop overhead, the browser/worker doesn't pause to schedule promises, resulting in immediate execution loops.
2. Zero-Copy Garbage Collection (Memory Fluidity)
Post-quantum cryptography is mathematically heavy and handles enormous matrices of bytes.
Version 0.7.1 optimized how Uint8Array buffers are recycled internally:
Fewer Array Allocations: The library now reuses memory arrays rather than constantly instantiating new intermediate arrays in RAM.Reduced GC Pauses: Because fewer temporary arrays are created, the browser's Garbage Collector (GC) doesn't have to pause execution to clean up memory junk, preventing micro-stutters during heavy encryption loops.
3. Loop Unrolling and Optimized NTTRef (Number Theoretic Transform)
Both ML-KEM and ML-DSA rely on polynomial multiplications handled via the Number Theoretic Transform (NTT). The math engine in 0.7.1 was tightly optimized:
Loop Unrolling: Critical mathematical loops were unrolled to execute consecutive assembly-like operations rather than running thousands of generic iterations.
Bitwise Efficiency: Core operations leverage native JavaScript 32-bit integer boundaries more efficiently, which modern V8 engines (Chrome, Edge, Node.js) optimize flawlessly.
4. Leaner Tree-ShakingBy modularizing the library paths (@noble/post-quantum/ml-kem.js), the bundle compiled by esbuild strips away all unneeded cryptographic overhead (like Falcon or Sphincs+ parameters). This ensures the CPU cache only deals with the lean, targeted binary instructions of your two specific modules.