<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Arkadiusz Przychocki on foojay.io - Friends of OpenJDK</title><link>https://foojayio.github.io/website/today/author/arkadiusz-przychocki/</link><description>Articles written by Arkadiusz Przychocki on foojay.io - Friends of OpenJDK</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Wed, 10 Jun 2026 12:00:28 +0000</lastBuildDate><atom:link href="https://foojayio.github.io/website/today/author/arkadiusz-przychocki/index.xml" rel="self" type="application/rss+xml"/><item><title>Your TLS Stack Is Lying to You About Zero-Copy</title><link>https://foojayio.github.io/website/today/your-tls-stack-is-lying-about-zero-copy/</link><pubDate>Wed, 10 Jun 2026 12:00:28 +0000</pubDate><guid>https://foojayio.github.io/website/today/your-tls-stack-is-lying-about-zero-copy/</guid><description>&lt;h2 id="h2-0-the-no-waste-compute-constraint"&gt;The &amp;ldquo;No Waste Compute&amp;rdquo; Constraint&lt;/h2&gt;
&lt;p&gt;When I started designing the Exeris Kernel, I set one non-negotiable rule very early: no waste compute. That rule sounds like a performance slogan until it starts killing otherwise normal design decisions.&lt;/p&gt;
&lt;p&gt;I had already banned &lt;code&gt;ThreadLocal&lt;/code&gt;, moved context propagation to Scoped Values, and pushed more of the runtime into explicit off-heap ownership. The idea was simple: if the hot path is supposed to stay outside GC pressure, then memory shape and lifetime cannot be treated as incidental details.&lt;/p&gt;</description></item><item><title>Scoped Values: Why I Banned ThreadLocal in Exeris</title><link>https://foojayio.github.io/website/today/banned-threadlocal-java-scoped-values/</link><pubDate>Thu, 28 May 2026 09:36:00 +0000</pubDate><guid>https://foojayio.github.io/website/today/banned-threadlocal-java-scoped-values/</guid><description>&lt;h4 id="in-a-zero-copy-runtime-designed-for-1-vt-per-stream-density-threadlocal-is-a-performance-serial-killer-here-is-the-forensic-analysis-and-how-jep-506-scoped-values-changed-everything"&gt;In a zero-copy runtime designed for 1-VT-per-Stream density, ThreadLocal is a performance serial killer. Here is the forensic analysis and how JEP 506 Scoped Values changed everything.&lt;/h4&gt;
&lt;hr&gt;
&lt;p&gt;When I started designing the &lt;strong&gt;Exeris Kernel&lt;/strong&gt; &amp;mdash; a next-generation, zero-copy runtime built for Java 26+ &amp;mdash; I established one non-negotiable architectural law: &lt;strong&gt;&amp;ldquo;No Waste Compute.&amp;rdquo;&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;In a system designed to handle extreme density by mapping exactly one Virtual Thread to every network stream (1-VT-per-Stream), every byte of memory and every CPU cycle must be intentional.&lt;/p&gt;</description></item></channel></rss>