<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Rust on miti99</title><link>https://miti99.com/en/tags/rust/</link><description>Recent content in Rust on miti99</description><generator>Hugo</generator><language>en</language><lastBuildDate>Mon, 16 Nov 2026 17:39:19 +0700</lastBuildDate><atom:link href="https://miti99.com/en/tags/rust/index.xml" rel="self" type="application/rss+xml"/><item><title>Newsletter #113</title><link>https://miti99.com/en/post/2026/06/24/</link><pubDate>Wed, 24 Jun 2026 00:00:00 +0700</pubDate><guid>https://miti99.com/en/post/2026/06/24/</guid><description>&lt;p&gt;&lt;em&gt;Enjoy Newsletter #113.&lt;/em&gt;&lt;/p&gt;
&lt;h2 id="migrating-from-go-to-rust"&gt;&lt;a href="https://corrode.dev/learn/migration-guides/go-to-rust/" rel="noopener noreferrer"&gt;Migrating from Go to Rust&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Matthias Endler writes a comprehensive guide for backend teams considering migrating services from Go to Rust. The central message isn&amp;rsquo;t &amp;ldquo;Rust is always faster than Go,&amp;rdquo; but that Rust pushes many classes of risk into the type system so the compiler catches them before production: &lt;code&gt;nil&lt;/code&gt; values, error handling, data races, resource lifetimes, and ownership. Go optimizes for team velocity—fast compile times, a strong standard library, and easy-to-use goroutines. Rust trades that for a stricter compiler, no GC, &lt;code&gt;Result&lt;/code&gt;, &lt;code&gt;Option&lt;/code&gt;, traits, and &lt;code&gt;Send&lt;/code&gt;/&lt;code&gt;Sync&lt;/code&gt; checks. Familiar Go habits have to be remapped: &lt;code&gt;if err != nil&lt;/code&gt; becomes &lt;code&gt;Result&lt;/code&gt; with &lt;code&gt;?&lt;/code&gt;, &lt;code&gt;nil&lt;/code&gt; becomes &lt;code&gt;Option&lt;/code&gt;, interfaces become traits, and goroutines become tasks when genuinely needed.&lt;/p&gt;
&lt;p&gt;The most practical section covers migration strategy: don&amp;rsquo;t rewrite everything at once. Start with a hot path, worker, or clearly bounded service, keep the existing API contract, and shift traffic gradually endpoint by endpoint through a gateway using the strangler pattern. While &lt;code&gt;cgo&lt;/code&gt; works, it is often more complex than splitting the Rust portion into a separate service. The author is also candid about the costs: the borrow checker makes the first month rough, compile times are slower, async function coloring is annoying, and parts of the ecosystem are smaller. There is no need to abandon Go entirely: Go remains a great fit for Kubernetes tooling, CLIs, and &amp;ldquo;boring&amp;rdquo; glue services, while Rust is worth the investment for foundational services where reliability, P99 latency, and bugs like nil dereferences or data races carry higher operational costs than the learning and build overhead.&lt;/p&gt;
&lt;h2 id="ai-engineering-for-developers"&gt;&lt;a href="https://www.lucavallin.com/blog/ai-engineering-for-developers" rel="noopener noreferrer"&gt;AI Engineering for Developers&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Luca Cavallin writes a deep-dive guide to AI engineering for developers already comfortable with backends, HTTP, queues, Kubernetes, and production operations. His main argument: when bringing LLMs into a product, the model is not the whole product; what actually creates value is the system around it—prompts, RAG, evals, agents, inference, observability, security, and cost controls. That makes the discipline much closer to backend engineering than model training. Because AI outputs are non-deterministic and can shift across models, prompts, or data, a working demo isn&amp;rsquo;t enough to ship: you need eval datasets, regression checks on every prompt or model change, canary rollouts, and monitoring from day one.&lt;/p&gt;
&lt;p&gt;The article progresses from fundamentals to practical execution: selecting models by task, cost, and latency; exhausting prompt engineering before reaching for RAG; treating RAG primarily as a search problem where hybrid search, rerankers, chunking strategy, and source citations matter more than stuffing more context; fine-tuning only after prompting and RAG hit their ceiling; and optimizing inference through caching, batching, and model routing. Agents are framed as tool-calling loops that require strict schemas, narrow permissions, end-to-end tracing, and human approval gates for risky actions. The production section emphasizes guardrails, dedicated CI/CD pipelines for prompt and model changes, per-feature cost attribution, and treating every agent as a service with its own identity, least-privilege access, and audit logs.&lt;/p&gt;
&lt;h2 id="the-invisible-engineering-behind-lambdas-network"&gt;&lt;a href="https://www.allthingsdistributed.com/2026/04/the-invisible-engineering-behind-lambdas-network.html" rel="noopener noreferrer"&gt;The invisible engineering behind Lambda’s network&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;This post on All Things Distributed recounts nearly a decade of networking optimizations behind AWS Lambda—the kind of &amp;ldquo;invisible&amp;rdquo; engineering customers only notice because functions start faster and more reliably. The original bottleneck was cold starts when Lambda connected to a customer VPC: beyond spinning up a Firecracker microVM and initializing the runtime, the system had to establish a private network path into the customer&amp;rsquo;s VPC, with Geneve tunnel creation sitting right on the hot path. Rather than maintaining custom kernel patches, the Lambda team turned to eBPF: pre-creating tunnels with a dummy VNI and overwriting the packet header once the real VNI arrived, cutting tunnel setup latency from ~150ms to 200 microseconds.&lt;/p&gt;
&lt;p&gt;Even more instructive is how they scaled to 4,000 networks per worker. Instead of creating tap devices, veth pairs, namespaces, and routing rules on every function invocation, they pre-allocated everything at worker startup, turning a load-variable cost into a one-time fixed cost. They replaced stateful NAT with stateless eBPF packet rewriting, cutting NAT setup latency 100x; collapsed over 125,000 iptables rules in the root namespace down to 144 static rules by moving per-slot rules into their respective namespaces; and reduced RTNL lock contention by reordering network creation and attaching eBPF programs in batches. At scale, iptables rules, conntrack, or a single kernel lock can all become bottlenecks. The result is a unified topology serving both traditional workloads and SnapStart, boosting snapshot network throughput 20x and getting repackaged for reuse in Aurora DSQL.&lt;/p&gt;
&lt;h2 id="selective-test-execution-at-stripe-fast-ci-for-a-50m-line-ruby-monorepo"&gt;&lt;a href="https://stripe.dev/blog/selective-test-execution-at-stripe-fast-ci-for-a-50m-line-ruby-monorepo" rel="noopener noreferrer"&gt;Selective Test Execution at Stripe: Fast CI for a 50M-line Ruby monorepo&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Aditya Anchuri describes how Stripe keeps CI fast for a ~50-million-line Ruby monorepo containing nearly 100,000 test files and 1.2 million test targets. Running the entire suite sequentially would take roughly four months, while Stripe runs around 50,000 builds per week. Their Selective Test Execution (STE) system solves this by running an average of just ~5% of the test suite per build (with a median under 0.5%), consuming less than 10% of the compute required by a run-everything strategy.&lt;/p&gt;
&lt;p&gt;What makes Stripe&amp;rsquo;s approach clever is that they don&amp;rsquo;t attempt static dependency analysis for Ruby, where metaprogramming, dynamic dispatch, configuration files, and code generation make static analysis either miss dependencies or become overly conservative. Instead, they observe runtime behavior: an internal C++ library injected via &lt;code&gt;LD_PRELOAD&lt;/code&gt; intercepts every file open, associates read files with the currently executing test, and propagates tracking across child processes. The &lt;code&gt;open&lt;/code&gt; path writes only minimal logs, keeping aggregation and indexing off the hot path. Those logs are compiled into a roaring bitmap index mapping changed files to required tests, storing roughly three billion data points while remaining fast to query. File changes are detected using &lt;code&gt;hashdeep&lt;/code&gt; so generated files are covered alongside git-tracked files. STE also relies on reproducible baseline data and practical CI safety nets: always re-running currently failing tests, always running tests that glob the filesystem, handling linters via changed-file lists, and storing metadata in MongoDB with Monotonic Revision IDs to quickly select baselines in commit order.&lt;/p&gt;
&lt;h2 id="being-oncall-taught-me-everything"&gt;&lt;a href="https://yaoyue.org/blog/2026-oncall/" rel="noopener noreferrer"&gt;Being oncall taught me everything&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Yao Yue looks back on 7.5 years of being on call for Twitter&amp;rsquo;s distributed caching layer—the highest-throughput service and the one involved in the most severe site-wide outages. The post doesn&amp;rsquo;t romanticize on-call life, but argues that living alongside production is what shaped the author&amp;rsquo;s mindset as an infrastructure engineer. On-call teaches you that at scale, predictability matters more than being fast in the average case, tail latency matters more than the median, simple and clear architecture is a lifeline during a crisis, and operational excellence (deep observability, consistent configuration, automation readiness, sensible defaults) comes from up-front design rather than frantic patching two weeks before launch.&lt;/p&gt;
&lt;p&gt;Even more memorable are the human lessons. Periodically restarting Memcached to prevent site-wide cascading failures sometimes triggered incidents of its own, teaching the author to pick low-risk windows, give advance notice, and—most importantly—own mistakes immediately, since that is the fastest way to earn back teammates&amp;rsquo; trust. Many outages were resolved only through doggedly narrowing down possibilities and leaning on colleagues across every layer of the stack: site ops, kernel engineers, upstream service owners, managers, and fellow on-call engineers. The conclusion is blunt: engineers don&amp;rsquo;t truly understand software until they see it run and break in production, and people who write software shouldn&amp;rsquo;t consider themselves above deploying, monitoring, or debugging what they build.&lt;/p&gt;
&lt;h2 id="claude-as-your-performance-analysis-partner"&gt;&lt;a href="https://developers.redhat.com/articles/2026/05/29/claude-your-performance-analysis-partner" rel="noopener noreferrer"&gt;Claude as your performance analysis partner&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Archana Ravindar experiments with using Claude as a partner for Go performance analysis across CPU profiles and execution traces, focusing on the Green Tea garbage collector on POWER10 using the &lt;code&gt;sweet&lt;/code&gt; benchmark suite. Claude&amp;rsquo;s value doesn&amp;rsquo;t lie in immediately generating a working optimization patch, but in rapidly digesting massive, dense artifacts like &lt;code&gt;pprof&lt;/code&gt; outputs, assembly dumps, and trace timelines. For example, Claude pinpointed a hot path in &lt;code&gt;runtime.tryDeferToSpanScan&lt;/code&gt;, spotted an atomic &lt;code&gt;Load8&lt;/code&gt; followed by &lt;code&gt;Or8&lt;/code&gt; pattern, and suggested fusing them into a single &lt;code&gt;Or32&lt;/code&gt; instruction. That experiment actually ran slower, and Claude then helped explain why: false sharing and contention across GC workers—a clear illustration that fewer atomic instructions doesn&amp;rsquo;t automatically mean faster code.&lt;/p&gt;
&lt;p&gt;Claude also proved useful for low-level analysis: cross-referencing assembly to see that the compiler wasn&amp;rsquo;t caching &lt;code&gt;q.class.sizeclass&lt;/code&gt;, ranking GC TODO comments by profile heat, and suggesting patterns to look for in traces such as idle processors during GC, mark assist gaps, overly frequent GC cycles, or prolonged &lt;code&gt;gcMarkDone&lt;/code&gt; phases. The author draws a clear distinction: profiles tell you where cost lives, while traces explain why the system is stalling, especially around GC and scheduling. Still, every hypothesis must be benchmarked, because optimizations tied to hardware architecture, cache lines, instruction sets, or register pressure are easy to get wrong without full context; the best use of AI here is narrowing the search space and building an investigation checklist, not replacing benchmarks.&lt;/p&gt;
&lt;h2 id="things-you-didn"&gt;&lt;a href="https://jon.chrt.dev/2026/04/15/things-you-didnt-know-about-indexes.html" rel="noopener noreferrer"&gt;Things you didn&amp;rsquo;t know about indexes&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Jon Charter explains Postgres indexes with an approachable analogy: an index is like the index at the back of a book, letting the database locate rows through a sorted structure instead of scanning the entire table. That speed comes with a fundamental trade-off: faster reads mean slower writes. Data modifications (&lt;code&gt;INSERT&lt;/code&gt;, &lt;code&gt;UPDATE&lt;/code&gt;, &lt;code&gt;DELETE&lt;/code&gt;) usually have to update indexes as well and incur cleanup overhead; indexes also consume disk space, compete for cache memory, and give the query planner more paths to evaluate. Consequently, &amp;ldquo;index everything&amp;rdquo; is rarely a good strategy.&lt;/p&gt;
&lt;p&gt;The most useful section covers common pitfalls that prevent an index from being used. Composite indexes depend on column order, so the order must match real query patterns: &lt;code&gt;(type_1, type_2)&lt;/code&gt; helps queries filtering on &lt;code&gt;type_1&lt;/code&gt; or both columns, but is generally ineffective when filtering only on &lt;code&gt;type_2&lt;/code&gt; (though PostgreSQL 18&amp;rsquo;s skip scan helps in some cases). Wrapping a column in a function like &lt;code&gt;lower(name)&lt;/code&gt; causes a plain index on &lt;code&gt;name&lt;/code&gt; to be ignored because the database needs a structure sorted by the exact expression &lt;code&gt;lower(name)&lt;/code&gt;; implicit type casts cause the same issue. The right way to verify index usage is running &lt;code&gt;EXPLAIN&lt;/code&gt; or &lt;code&gt;EXPLAIN ANALYZE&lt;/code&gt; to confirm what the planner actually chooses rather than guessing. Finally, the post introduces functional indexes, partial indexes for small slices of data like soft-deleted rows or &lt;code&gt;is_legendary = true&lt;/code&gt;, and covering indexes with &lt;code&gt;INCLUDE&lt;/code&gt; to enable &lt;code&gt;Index Only Scan&lt;/code&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;&lt;em&gt;This post was rewritten by Claude Code with Opus 5.5 on 27/09/2026; reviewed and edited by Codex with GPT 6.1 Sol on 09/11/2026, and re-verified by Claude Code with Sonnet 5.5 on 16/11/2026.&lt;/em&gt;&lt;/p&gt;</description></item><item><title>Newsletter #112</title><link>https://miti99.com/en/post/2026/06/23/</link><pubDate>Tue, 23 Jun 2026 00:00:00 +0700</pubDate><guid>https://miti99.com/en/post/2026/06/23/</guid><description>&lt;p&gt;&lt;em&gt;Enjoy Newsletter #112.&lt;/em&gt;&lt;/p&gt;
&lt;h2 id="a-few-interesting-modern-pixel-fonts"&gt;&lt;a href="https://unsung.aresluna.org/a-few-interesting-modern-pixel-fonts/" rel="noopener noreferrer"&gt;A few interesting modern pixel fonts&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Marcin Wichary surveys a few modern pixel fonts built around interesting ideas. Andrew Gleeson&amp;rsquo;s Analog Mono &amp;ldquo;atones for the sins&amp;rdquo; of VCR OSD Mono, the classic pixel font seen across 1990s VCRs, TVs, and camcorders whose baseline sat so low that every character with a descender was pulled upward. Kumiko Yoshida&amp;rsquo;s Coral Pixels is a color font that deliberately preserves the characteristic color fringing of 1990s–2000s subpixel rendering: what used to be a technical artifact is now nostalgic material and an intentional visual element. Joseph Fatula&amp;rsquo;s Two Slice pushes minimalism to the extreme at just 2 pixels tall while remaining &amp;ldquo;barely readable.&amp;rdquo;&lt;/p&gt;
&lt;p&gt;The author notes that all of these are vector fonts masquerading as pixels so they can be installed on modern operating systems, which leads to Vercel&amp;rsquo;s Geist Pixel. Although its marketing pitch is slightly grandiose, it hits on an important problem: pixel fonts often break when put into real products because they don&amp;rsquo;t scale well across screen sizes, their metrics clash with existing typography systems, or they are purely decorative. According to the author, the Achilles&amp;rsquo; heel of many fonts lies not in the letterforms themselves, but in the &amp;ldquo;invisible&amp;rdquo; work around them such as kerning, metadata, extended character sets, and vertical metrics. The lesson for UI designers is that nostalgia can be a starting point, but integration stability is what determines whether a font is actually usable in a product.&lt;/p&gt;
&lt;h2 id="the-best-engineers-write-less-code"&gt;&lt;a href="https://shvetsm.github.io/posts/the-best-engineers-write-less-code/" rel="noopener noreferrer"&gt;The Best Engineers Write Less Code&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;The author pushes back against the idea that we no longer need to write good software because LLMs can generate code. Bad software is still bad software, whether written by Claude Code or an inexperienced developer: latency, complexity, and maintenance costs remain real, and there will be plenty of work cleaning up the low-quality code currently being pushed to production. Businesses measure everything in time, because time is money; that is also why they are excited about AI—it is faster and cheaper.&lt;/p&gt;
&lt;p&gt;According to the author, the best thing an engineer can do for stakeholders is avoid building unnecessary things, since every feature carries long-term costs in maintenance, debugging, infrastructure, support, migrations, and cognitive load. On the path to becoming a senior engineer, you shift from &amp;ldquo;I can build that, it&amp;rsquo;ll take a few days&amp;rdquo; to &amp;ldquo;Why are we building this?&amp;rdquo;, and the second question is critical: code is a liability unless it solves the right problem. The worst outcome is building the wrong thing. The way to reduce risk is genuine Agile in spirit rather than ceremony: align on why the work is needed, adjust requirements to deliver value faster, and work in short intervals to show progress and gather feedback. Great engineers are measured not by the volume of code they produce, but by the value they deliver relative to the complexity they leave behind—and the best use of spare time is deleting code.&lt;/p&gt;
&lt;h2 id="fast-is-better-than-slow"&gt;&lt;a href="https://dubroy.com/blog/fast-is-better-than-slow/" rel="noopener noreferrer"&gt;Fast is better than slow&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Patrick Dubroy noticed that the best programmers he has ever worked with all share one trait: they are fast, often turning up with a patch or prototype an hour or two after discussing a problem. He argues that they aren&amp;rsquo;t fast because they&amp;rsquo;re good, they&amp;rsquo;re good because they&amp;rsquo;re fast: moving quickly gets you data sooner, leads to better decisions, compounds learning over time, and lets you try multiple approaches before picking the best one. This does not mean hustle culture or working overtime; there are many ways to speed up without putting in extra hours.&lt;/p&gt;
&lt;p&gt;His advice has more to do with day-to-day software engineering habits than raw coding skill. Don&amp;rsquo;t procrastinate just because you aren&amp;rsquo;t sure where to start, as waiting rarely makes the task easier. Make use of small pockets of time instead of waiting for long uninterrupted blocks, since fragmentation can eat up 25–35% of your day. Don&amp;rsquo;t be afraid to look silly: share work early using the &amp;ldquo;70% rule,&amp;rdquo; open a PR right away and welcome feedback instead of polishing endlessly, and proactively ask coworkers for help because software development is a team sport. Pick your battles: in code review, most disagreements are small enough that making the requested change is faster than arguing, and Dubroy himself often approves with &amp;ldquo;LGTM with comments.&amp;rdquo; Finally, build only what was asked for—trying to &amp;ldquo;overdeliver&amp;rdquo; usually means guessing at future requirements, and the less experience you have, the more likely you are to guess wrong.&lt;/p&gt;
&lt;h2 id="java-cheat-sheet"&gt;&lt;a href="https://tms-outsource.com/cs/java-cheat-sheet/" rel="noopener noreferrer"&gt;Java Cheat Sheet&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;This is a quick Java reference by Bogdan Sandu focused on syntax, common patterns, and immediately applicable idioms. The material is organized into sections: data types and casting, operators, control flow, arrays, strings and &lt;code&gt;StringBuilder&lt;/code&gt;, classes and objects, OOP concepts, access modifiers, the Collections Framework, exception handling, streams and lambdas, concurrency, I/O, and modern Java features. Helpfully, it doesn&amp;rsquo;t just list syntax—it also calls out classic beginner pitfalls like comparing &lt;code&gt;String&lt;/code&gt; instances with &lt;code&gt;==&lt;/code&gt;, confusing array &lt;code&gt;.length&lt;/code&gt; with string &lt;code&gt;.length()&lt;/code&gt;, handling &lt;code&gt;null&lt;/code&gt;, going out of array bounds, or concatenating strings in a loop instead of using &lt;code&gt;StringBuilder&lt;/code&gt;.&lt;/p&gt;
&lt;p&gt;For junior developers, a cheat sheet like this is well suited for a quick refresher when switching to Java from another language, or when recalling a standard API without wading through long documentation. Even so, it should be treated as an initial lookup map: it does not replace official docs, a deep understanding of the JVM, OOP, collections, and concurrency, or hands-on experience writing production code.&lt;/p&gt;
&lt;h2 id="swapping-go-for-rust-10x-cheaper-k8s-ingress"&gt;&lt;a href="https://dev.to/syedahmershah/swapping-go-for-rust-10x-cheaper-k8s-ingress-2b4p" rel="noopener noreferrer"&gt;Swapping Go for Rust: 10x Cheaper K8s Ingress&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Syed Ahmer Shah recounts cutting the AWS bill for a Kubernetes ingress layer from roughly $4,200 to about $390 per month. The original setup ran Traefik across 3 pods, averaging around 380MB RSS per pod (roughly 180MB idle, over 400MB under load), requiring &lt;code&gt;t3.large&lt;/code&gt; nodes for headroom. The author attributes this to Go&amp;rsquo;s garbage collector, which optimizes for latency rather than memory footprint: in a proxy, every request allocates headers, buffers, and connection state, so under sustained load the GC rarely gets a &amp;ldquo;good moment&amp;rdquo; to clean up, whereas Rust frees memory deterministically as soon as values go out of scope. It is worth noting that this changed both the architecture and infrastructure (Envoy, smaller nodes), so it cannot be treated as an isolated benchmark of swapping languages alone.&lt;/p&gt;
&lt;p&gt;Rather than rewriting Traefik, the team moved core proxying to Envoy (C++) alongside a small Rust service that took over custom routing logic previously living in middleware. The first week hit three incidents—not because of Rust, but because Envoy&amp;rsquo;s explicit configuration model differed sharply from Traefik&amp;rsquo;s annotation &amp;ldquo;magic.&amp;rdquo; After about three weeks, the ingress layer dropped from 3 Traefik pods at ~380MB RSS each to 2 Envoy pods plus the Rust layer at ~40MB per pod; node costs fell from roughly $340 to $30 per month. The takeaway isn&amp;rsquo;t &amp;ldquo;replace Go with Rust&amp;rdquo;: Go remains a great fit for APIs, CLIs, and standard microservices, while Rust is worth considering for high-performance data planes. The migration took two engineers about three weeks, so if Traefik is working fine and the bill isn&amp;rsquo;t hurting, leave it alone.&lt;/p&gt;
&lt;h2 id="persistent-multiplayer-state-without-chaos"&gt;&lt;a href="https://packagemain.tech/p/persistent-multiplayer-state-without" rel="noopener noreferrer"&gt;Persistent multiplayer state without chaos&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Julien Singler shares an architecture for maintaining consistent state in a live multiplayer game, where PostgreSQL holds the canonical data and Redis handles speed-critical operations. In a single-player game, state is relatively straightforward; but when multiple players can act on the same target simultaneously, scheduled jobs must finish on time regardless of who is online, and players returning days later expect their progress intact, state becomes the hardest problem. Keeping critical data only in memory or flushing to disk only on logout easily leads to lost progress or economy bugs.&lt;/p&gt;
&lt;p&gt;The Go backend is structured into handler, service, and repository layers. Every critical state change goes through a PostgreSQL transaction as the single source of truth. Read-heavy data is fetched from Redis first, falling back to PostgreSQL on a cache miss, and the cache is invalidated on writes before returning success so players never see stale state. Redis holds only rebuildable state such as caches, online presence, and short-lived locks: distributed locks via &lt;code&gt;SetNX&lt;/code&gt; ensure periodic jobs run only once across multiple backend instances (though bare &lt;code&gt;SetNX&lt;/code&gt; alone does not guarantee exactly-once processing if a crash or lock expiration occurs). Scheduled jobs rely on an indexed &lt;code&gt;finishes_at&lt;/code&gt; column rather than scanning every player every minute, and per-player mutations use &lt;code&gt;SELECT ... FOR UPDATE&lt;/code&gt; inside transactions instead of a custom lock manager.&lt;/p&gt;
&lt;h2 id="we"&gt;&lt;a href="https://plc.vc/cdx" rel="noopener noreferrer"&gt;We&amp;rsquo;re in the Over-engineering Game Now&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Peter Clark writes that he spent his entire career priding himself on scoping projects to minimize engineering effort while maximizing flexibility, since engineering time is a software company&amp;rsquo;s scarcest resource. Yet he also loves product design and polished details, leaving him constantly torn between wanting to craft things more carefully and worrying about wasting resources. Now, using Claude and Codex on personal projects, he happily &amp;ldquo;over-engineers&amp;rdquo;: his blog has a dynamic changelog, a &amp;ldquo;vibe meter,&amp;rdquo; an MCP that generates weekly roundups, and hundreds of other ideas he tried and threw away.&lt;/p&gt;
&lt;p&gt;His central thesis is that AI coding blurs the line between over-engineering and simply finishing a product well. His iOS app no longer starts as a bare-bones MVP, but launches feature-complete with built-in feedback loops that trigger agentic workflows. Improvements that used to languish forever in the backlog because they were never high enough priority—or too small to even file a ticket for—can now be iterated on and polished continuously. He argues this shift will raise user expectations: when building good software becomes nearly free, clunky products will be much harder to excuse. Rather than merely an era of vibe coding or apps for an audience of one, he believes we are entering a phase where product vision is no longer bottlenecked by engineering resources, leaving only one question: what do you actually want to build?&lt;/p&gt;
&lt;h2 id="package-managers-that-package-package-managers"&gt;&lt;a href="https://nesbitt.io/2026/05/28/package-managers-that-package-package-managers.html" rel="noopener noreferrer"&gt;Package managers that package package managers&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Andrew Nesbitt built a matrix mapping which package managers can install which other package managers. The idea started from a strange loop: PyPI has a package containing the Node binary, and npm has a package containing a portable CPython build, meaning &lt;code&gt;pip install&lt;/code&gt; and &lt;code&gt;npm install&lt;/code&gt; can hand control back and forth indefinitely. Covering 42 clients using data from ecosyste.ms for language registries and Repology for Linux distributions, the densest rows belong to system package managers like AUR (40/42), nixpkgs, Homebrew, DNF, and Debian, since packaging arbitrary binaries is their job; as columns, however, they are nearly empty because nobody needs to redistribute &lt;code&gt;apt&lt;/code&gt; or DNF when they ship with the OS. Conda sits between the two worlds, while PyPI is the most prolific language registry because cross-language tools like Conan and Meson are written in Python.&lt;/p&gt;
&lt;p&gt;The post also highlights notable software supply chain implications. Twenty-five of the 42 tools publish themselves through their own registry. A CVE in pip might be tracked under &lt;code&gt;pkg:pypi/pip&lt;/code&gt; or &lt;code&gt;pkg:deb/python3-pip&lt;/code&gt;, but copies of pip inside Homebrew, conda-forge, nixpkgs, or Spack carry the same vulnerability and each ecosystem must map advisories on its own, making upstream vulnerability tracing much harder. Searching by package name is also unreliable because short names on flat-namespace registries are often squatted by unrelated projects, whereas reverse-linking to the source repository yields cleaner matches. Finally, the author constructs a 14-step installation chain stretching from an Arch Linux machine to the Elm compiler.&lt;/p&gt;
&lt;h2 id="beyond-the-prompt-claude-code"&gt;&lt;a href="https://arps18.github.io/posts/claude-code-mastery/" rel="noopener noreferrer"&gt;Beyond the Prompt: Claude Code&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Arpan Patel writes an extensive guide to using Claude Code as a workflow-driven agent rather than a prompt-in, code-out chatbot. The core focus is building feedback loops so Claude can verify its own work: reading code, planning, editing, running tests, reviewing, and updating its rules whenever it makes a mistake. The author recommends exploring before coding, using plan mode for multi-file changes, pointing directly to files or logs instead of giving vague descriptions, and delegating tasks the way you would to an engineer rather than micromanaging every small step.&lt;/p&gt;
&lt;p&gt;A standout section covers organizing the &lt;code&gt;.claude&lt;/code&gt; directory: &lt;code&gt;CLAUDE.md&lt;/code&gt; for shared project rules, &lt;code&gt;CLAUDE.local.md&lt;/code&gt; for personal notes, alongside settings, MCP servers, skills, commands, rules, and subagents. &lt;code&gt;CLAUDE.md&lt;/code&gt; should stay concise like a guardrail—containing only rules that genuinely prevent mistakes rather than documenting the entire codebase—while recurring workflows should be turned into reusable skills, commands, or subagents in the repo. For larger teams, plugins, MCPs, and subagents help isolate context, run independent reviews, and execute parallel streams of work, delivering the most value when baked into shared onboarding rather than kept as personal tweaks.&lt;/p&gt;
&lt;h2 id="how"&gt;&lt;a href="https://performance.dev/how-is-linear-so-fast-a-technical-breakdown" rel="noopener noreferrer"&gt;How&amp;rsquo;s Linear so fast? A technical breakdown&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Dennis Brotzky breaks down why Linear feels instantaneous despite being a client-side web application. The core idea is that Linear removes network latency from the primary interaction path. The data the UI reads is already in the browser, primarily via IndexedDB and an in-memory object pool; mutations are applied optimistically locally first, written to a persistent transaction queue, and then synced to the server. The server acts as an arbiter that acknowledges changes and broadcasts deltas back, rather than something the UI waits on before responding to the user. At runtime, MobX enables fine-grained updates so each delta re-renders only the specific fields and components affected rather than an entire list.&lt;/p&gt;
&lt;p&gt;The post also dives into initial load performance, where speed comes from compounding small details: trimming JavaScript and CSS payloads, targeting only modern browsers, aggressive code splitting, preloading modules to avoid waterfall requests, service worker caching, preloading fonts with the proper &lt;code&gt;crossorigin&lt;/code&gt; attribute, embedding an app shell, and rendering local data before background authentication finishes. Beyond engineering, speed comes from product design: every action has a keyboard shortcut, the command palette is always one keystroke away and runs against local data, and animations are brief and stick to cheap properties like &lt;code&gt;transform&lt;/code&gt; and &lt;code&gt;opacity&lt;/code&gt;. Performance is thus as much about user perception as it is about raw engineering.&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;&lt;em&gt;This post was rewritten by Claude Code with Opus 5.5 on 27/09/2026; reviewed and edited by Codex with GPT 6.1 Sol on 09/11/2026, and re-verified by Claude Code with Sonnet 5.5 on 16/11/2026.&lt;/em&gt;&lt;/p&gt;</description></item><item><title>Newsletter #111</title><link>https://miti99.com/en/post/2026/06/11/</link><pubDate>Thu, 11 Jun 2026 00:00:00 +0700</pubDate><guid>https://miti99.com/en/post/2026/06/11/</guid><description>&lt;p&gt;&lt;em&gt;Enjoy Newsletter #111.&lt;/em&gt;&lt;/p&gt;
&lt;h2 id="learnings-from-100k-lines-of-rust-with-ai"&gt;&lt;a href="https://zfhuang99.github.io/rust/claude%20code/codex/contracts/spec-driven%20development/2025/12/01/rust-with-ai.html" rel="noopener noreferrer"&gt;Learnings from 100K Lines of Rust with AI&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Cheng Huang recounts using AI agents like Claude Code, Codex CLI, and Copilot to modernize Azure&amp;rsquo;s Replicated State Library (RSL) — a multi-Paxos consensus engine — in Rust. The resulting codebase reached roughly 130k lines of code and over 1,300 tests. To get AI to correctly implement such a complex distributed system, the author relied heavily on &amp;ldquo;code contracts&amp;rdquo;: having the AI write preconditions, postconditions, and invariants for critical functions, enforcing them as runtime &lt;code&gt;assert&lt;/code&gt;s during testing, and using them to drive both targeted unit tests and property-based tests. That approach caught a subtle Paxos safety violation before the code ever reached production.&lt;/p&gt;
&lt;p&gt;The second technique was lightweight spec-driven development: instead of maintaining rigid, heavyweight requirements and design docs, the author used &lt;code&gt;spec-kit&lt;/code&gt;&amp;rsquo;s &lt;code&gt;/specify&lt;/code&gt; command to generate user stories with acceptance criteria, followed by &lt;code&gt;/clarify&lt;/code&gt; so the AI could critique and refine them; each user story then became a right-sized unit of work for an agent. The third technique was AI-driven performance optimization: over roughly three weeks, the AI added latency instrumentation, ran benchmarks, analyzed bottlenecks, and proposed improvements like reducing allocations, zero-copy parsing, and cutting lock contention — pushing throughput from ~23k to 300k operations per second while Rust preserved memory safety. The author closes with the hope that agents will soon execute entire user stories end to end, automate the contract workflow, and optimize performance across even larger codebases, while humans retain architectural oversight.&lt;/p&gt;
&lt;h2 id="always-be-blaming"&gt;&lt;a href="https://matklad.github.io/2026/05/18/always-be-blaming.html" rel="noopener noreferrer"&gt;Always Be Blaming&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;matklad proposes reading code across multiple &amp;ldquo;dimensions.&amp;rdquo; 2D reading means forming your own mental sketch of how a solution should look first, then comparing it against the actual code to spot surprises; 3D reading means tracing how the code evolved over time, because most production code is Markovian — its shape today depends on its shape yesterday, not just on the problem it solves. The fourth dimension, and the highest goal, is reconstructing the original author&amp;rsquo;s reasoning and train of thought — applying &amp;ldquo;theory of mind&amp;rdquo; to reading code.&lt;/p&gt;
&lt;p&gt;To do that, matklad treats &lt;code&gt;git&lt;/code&gt; history as a first-class investigative tool. On GitHub, press &lt;code&gt;y&lt;/code&gt; to pin the URL to a specific commit SHA, open the blame view, and repeatedly click &amp;ldquo;blame prior to change,&amp;rdquo; opening relevant commits in new tabs and browsing surrounding files at that exact point in time. Locally, he uses custom keybindings: &lt;code&gt;,b l&lt;/code&gt; to check out the commit that last touched the current line, &lt;code&gt;,b p&lt;/code&gt; to step back to the parent commit, &lt;code&gt;,b u&lt;/code&gt; to jump back to the previous inspection point, and &lt;code&gt;,b w&lt;/code&gt; to copy the GitHub link. This &amp;ldquo;in-place blame&amp;rdquo; workflow keeps the LSP, test runner, and search tools like &lt;code&gt;rg&lt;/code&gt; working normally as you travel through time. He also laments that code review comments — an invaluable source of context — remain locked inside third-party hosting databases rather than living inside the Git repository itself.&lt;/p&gt;
&lt;h2 id="scaling-postgresql-to-power-800-million-chatgpt-users"&gt;&lt;a href="https://openai.com/index/scaling-postgresql/" rel="noopener noreferrer"&gt;Scaling PostgreSQL to power 800 million ChatGPT users&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Bohan Zhang shares how OpenAI scaled PostgreSQL to serve ChatGPT and its API across roughly 800 million users. In just one year, PostgreSQL load grew more than 10x, yet the system still relies on a single Azure primary for all writes paired with nearly 50 read replicas spread across multiple regions. The biggest pressure point is write amplification: PostgreSQL&amp;rsquo;s MVCC copies an entire row on every update, amplifying both writes and reads while bloating tables and indexes. To protect the primary, OpenAI offloaded write-heavy, easily partitionable workloads to Azure Cosmos DB, banned adding new tables to the existing PostgreSQL cluster, and held off on sharding because it would require rewriting hundreds of application endpoints.&lt;/p&gt;
&lt;p&gt;Most overload incidents followed the exact same cascade: an upstream trigger — a wave of cache misses, an expensive join, or a write spike from a feature launch — pushed latency up, requests timed out, and client retries amplified the load into a death spiral. To break that chain, OpenAI optimized expensive queries, deployed PgBouncer for connection pooling, added cache locking to prevent cache-miss stampedes, isolated workloads by priority tier, enforced multi-layer rate limiting, capped schema migrations at a 5-second timeout, and throttled backfills. They are also testing cascading replication with Azure to add more read replicas without making the primary ship WAL to all of them directly. As a result, PostgreSQL handles millions of queries per second while keeping client-side p99 latency in the tens of milliseconds and maintaining five-nines availability.&lt;/p&gt;
&lt;h2 id="scaling-managed-agents-decoupling-the-brain-from-the-hands"&gt;&lt;a href="https://www.anthropic.com/engineering/managed-agents" rel="noopener noreferrer"&gt;Scaling Managed Agents: Decoupling the brain from the hands&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Anthropic explains the architecture behind Claude Managed Agents, their hosted service for running long-lived AI agents. The starting observation is that agent harnesses often bake in workarounds for whatever current models struggle with, and those assumptions quickly go stale as models improve. Rather than bundling the model, harness, sandbox, and session state inside a single container, Anthropic decoupled them into three components behind stable interfaces — much like an operating system virtualizes hardware: the session is an append-only event log, the harness is the control loop coordinating Claude and tool calls, and the sandbox is the code execution environment. The pieces communicate through clean primitives like &lt;code&gt;execute()&lt;/code&gt;, &lt;code&gt;emitEvent()&lt;/code&gt;, &lt;code&gt;getSession()&lt;/code&gt;, and &lt;code&gt;getEvents()&lt;/code&gt;.&lt;/p&gt;
&lt;p&gt;This decoupling turns containers from &amp;ldquo;pets&amp;rdquo; that must be kept alive into disposable &amp;ldquo;cattle&amp;rdquo;: a crashed sandbox is just a failed tool call that can be spun up fresh, while the stateless harness can restart at any time and resume from the session event log. A single harness can also coordinate multiple sandboxes, including reaching into a customer&amp;rsquo;s VPC without requiring network peering. On security, credentials never live inside the sandbox where untrusted code runs; tokens are either bound to resources at provisioning time or held in a vault behind an authenticating proxy, making them much harder for prompt injections to steal. Latency improved dramatically too, because inference starts immediately without waiting for a container to boot — cutting time-to-first-token by roughly 60% at p50 and over 90% at p95.&lt;/p&gt;
&lt;h3 id="bonus"&gt;Bonus&lt;/h3&gt;&lt;p&gt;&lt;strong&gt;Images:&lt;/strong&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!fap_!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc111a2c6-71bd-45a7-8938-e8bb4f772cbc_1378x1390.png" alt="How does Claude Code keep long sessions from running out of context?"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!VfjL!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F38f19060-e838-440a-971d-bdc1f2d3a167_1280x1546.jpeg" alt="How does a request actually travel through Claude Code?"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!cqW0!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2e0de719-d0ac-497e-8f9c-b9c841422db8_2484x3002.png" alt="Forward Proxy, Reverse Proxy, and API Gateway Explained"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!eLEi!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F52ccd4c2-dc53-4b9a-95b9-d87d6c353f88_2484x3002.png" alt="RAGs vs Agents"&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;&lt;em&gt;This post was rewritten by Claude Code with Opus 5.5 on 27/09/2026.&lt;/em&gt;&lt;/p&gt;</description></item><item><title>Newsletter #93</title><link>https://miti99.com/en/post/2026/03/28/</link><pubDate>Sat, 28 Mar 2026 00:00:00 +0700</pubDate><guid>https://miti99.com/en/post/2026/03/28/</guid><description>&lt;p&gt;&lt;em&gt;Enjoy Newsletter #93.&lt;/em&gt;&lt;/p&gt;
&lt;h2 id="lessons-from-building-claude-code-how-we-use-skills"&gt;&lt;a href="https://x.com/trq212/status/2033949937936085378" rel="noopener noreferrer"&gt;Lessons from Building Claude Code: How We Use Skills&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Thariq Shihipar, an engineer at Anthropic, distills lessons from running hundreds of internal skills in Claude Code. The first misconception he corrects is that skills are &amp;ldquo;just markdown files&amp;rdquo;: each skill is actually a directory that can bundle scripts, reference assets, and data for the agent to discover and execute, along with configuration options like dynamic hook registration. Surveying their internal catalog, the team found that effective skills fall into nine recurring categories: library and API references, product verification, data fetching and analysis, team workflow automation, code scaffolding, code quality enforcement, CI/CD and deployment, incident runbooks, and infrastructure operations. Great skills fit cleanly into one category; confusing ones straddle several.&lt;/p&gt;
&lt;p&gt;When authoring skills, he advises skipping obvious instructions and focusing on where you need to nudge Claude away from its default habits—a &amp;ldquo;Gotchas&amp;rdquo; section built up from real failure modes is consistently the highest-leverage part. The frontmatter &lt;code&gt;description&lt;/code&gt; isn&amp;rsquo;t a human summary; it tells the model &lt;em&gt;when&lt;/em&gt; to trigger the skill. Skills can persist per-user settings in &lt;code&gt;config.json&lt;/code&gt; (prompting via &lt;code&gt;AskUserQuestion&lt;/code&gt; when missing), keep lightweight &amp;ldquo;memory&amp;rdquo; in append-only logs or JSON files inside a stable data directory, and ship helper scripts so Claude spends turns composing tools rather than regenerating boilerplate. Finally, skills can reference one another by name, and the team tracks skill invocations via a &lt;code&gt;PreToolUse&lt;/code&gt; hook to spot popular skills and prune dead ones.&lt;/p&gt;
&lt;h2 id="the-future-of-software-engineering-with-anthropic"&gt;&lt;a href="https://www.akashbajwa.co/p/the-future-of-software-engineering" rel="noopener noreferrer"&gt;The Future Of Software Engineering with Anthropic&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Akash Bajwa summarizes a roundtable with Anthropic&amp;rsquo;s Ash Prabaker and engineering leaders from Stripe, NVIDIA, Microsoft, Google DeepMind, xAI, Apple, Scale AI, and OpenAI on how AI is reshaping software development. Claude Code started in late 2024 as a humble CLI built not for the models of the day, but for where model capabilities would be six to twelve months later—and spread organically through bottom-up utility rather than top-down mandates. A recurring theme is the recursive improvement loop: better coding tools help build better models, which in turn improve the tools; several companies already run pipelines that autonomously triage production bugs, cross-check against eval suites, and open fix PRs with minimal human intervention.&lt;/p&gt;
&lt;p&gt;Across engineering workflows, test-first development has become the default, human code review increasingly feels more like a bottleneck than a safety net, and inline code comments are making a comeback because future agent sessions rely on them. Human-written architectural context remains invaluable, whereas stale or auto-generated documentation actively misleads agents. In hiring, several leaders now prize engineers who experiment relentlessly at the frontier of tool capabilities over raw hand-coding speed. Open challenges include multi-hour autonomous agent runs that still preserve meaningful human oversight, large-scale context management, and the paradox that when AI makes building almost anything feasible, ruthless prioritization becomes harder than ever.&lt;/p&gt;
&lt;h2 id="rob-pike"&gt;&lt;a href="https://www.cs.unc.edu/~stotts/COMP590-059-f24/robsrules.html" rel="noopener noreferrer"&gt;Rob Pike&amp;rsquo;s 5 Rules of Programming&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Rob Pike, co-creator of Go and veteran of Bell Labs, distilled five pithy rules of programming included here in the course readings for COMP590 at UNC. Rules 1 and 2 address optimization: you cannot tell in advance where a program will spend its time because bottlenecks show up in surprising places, so don&amp;rsquo;t put in speed hacks until you&amp;rsquo;ve measured and proven that one part of the code overwhelms the rest. Together, they restate Tony Hoare&amp;rsquo;s famous maxim that &amp;ldquo;premature optimization is the root of all evil.&amp;rdquo;&lt;/p&gt;
&lt;p&gt;Rules 3 and 4 warn that fancy algorithms have big constant factors that make them slow when &lt;code&gt;n&lt;/code&gt; is small—and &lt;code&gt;n&lt;/code&gt; is usually small—and they are much buggier and harder to implement, so stick to simple algorithms and simple data structures; as Ken Thompson put it, &amp;ldquo;when in doubt, use brute force.&amp;rdquo; Rule 5 is the heart of the list: data dominates. Once you&amp;rsquo;ve chosen the right data structures and organized things well, the algorithms will almost always be self-evident; data structures, not algorithms, are central to programming.&lt;/p&gt;
&lt;h2 id="introduction-to-postgresql-indexes"&gt;&lt;a href="https://dlt.github.io/blog/posts/introduction-to-postgresql-indexes/" rel="noopener noreferrer"&gt;Introduction to PostgreSQL Indexes&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Dalto Curvelano explains the internals of PostgreSQL indexes for developers who have an intuitive grasp of indexes but want to understand how they work under the hood. The article starts with how rows live inside 8KB heap pages and why indexes cut down page reads: in a benchmark on a million-row table, a sequential scan takes ~265ms while an index lookup takes just 0.077ms. As a rule of thumb, an index only pays off when a query returns less than roughly 15–20% of the table; beyond that, the query planner usually prefers a sequential scan. Nor are indexes free: they consume disk space, slow down &lt;code&gt;INSERT&lt;/code&gt;/&lt;code&gt;UPDATE&lt;/code&gt;/&lt;code&gt;DELETE&lt;/code&gt; operations, occupy buffer cache memory, and give the query planner more paths to evaluate, so they shouldn&amp;rsquo;t be sprinkled everywhere.&lt;/p&gt;
&lt;p&gt;Most of the guide surveys PostgreSQL&amp;rsquo;s index types. B-Tree is the versatile default; starting in PostgreSQL 18, B-Tree skip scan allows composite indexes to be used even when a query omits the leading column. Hash indexes only support equality checks, but are more compact than B-Trees on wide values like UUIDs or URLs; BRIN indexes are tiny and shine on append-only, naturally ordered time-series tables; GIN handles full-text search, arrays, and &lt;code&gt;JSONB&lt;/code&gt;; and GiST/SP-GiST cover geometric and range data. The post also walks through partial indexes (indexing only a filtered subset of rows), covering indexes with &lt;code&gt;INCLUDE&lt;/code&gt; (enabling index-only scans when the visibility map allows skipping heap fetches), and expression indexes over computed values.&lt;/p&gt;
&lt;h2 id="using-rust-and-postgres-for-everything-patterns-learned-over-the-years"&gt;&lt;a href="https://kerkour.com/rust-postgres-everything" rel="noopener noreferrer"&gt;Using Rust and Postgres for everything: patterns learned over the years&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Sylvain Kerkour shares the architectural patterns he has refined from running Rust and PostgreSQL as the backbone for nearly every backend project—guided by a philosophy of picking boring, dependable tools to minimize operational cost and complexity. He opens with a real-world rewrite of a data-processing service from Go to Rust paired with a high-performance memory allocator: batch processing time dropped from roughly 30 minutes to under 5 minutes, RAM usage fell from 4GB to 512MB, and nil-pointer panics disappeared entirely.&lt;/p&gt;
&lt;p&gt;On the application side, he favors &lt;code&gt;sqlx&lt;/code&gt; over heavy ORMs for its simplicity, performance, and compile-time SQL verification macros, and batches writes in chunks of up to 10,000 rows using &lt;code&gt;UNNEST&lt;/code&gt; to avoid overwhelming the database. Even better, PostgreSQL replaces a sprawl of auxiliary infrastructure: &lt;code&gt;pg_try_advisory_lock()&lt;/code&gt; handles leader election (for instance, coordinating distributed CRON schedulers) without ZooKeeper or Redis; &lt;code&gt;UNLOGGED&lt;/code&gt; tables replace Redis for ephemeral caches by skipping the WAL for much faster writes at the cost of crash durability; and PostgreSQL serves as a rock-solid job queue when combining UUIDv7 primary keys (to avoid B-tree index fragmentation) with &lt;code&gt;FOR UPDATE SKIP LOCKED&lt;/code&gt; for worker polling.&lt;/p&gt;
&lt;h2 id="air-traffic-control-the-ibm-9020"&gt;&lt;a href="https://computer.rip/2026-01-17-air-traffic-control-9020.html" rel="noopener noreferrer"&gt;air traffic control: the IBM 9020&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;J. B. Crawford tells the story of the IBM 9020—the custom multiprocessor mainframe built for the US Federal Aviation Administration (FAA) to automate en-route air traffic control, with the first full system installed in 1967. Earlier, the military&amp;rsquo;s SAGE air-defense computer had been considered for civilian traffic control, but it couldn&amp;rsquo;t verify whether assigned altitudes were unique or detect loss of safe separation between aircraft at a time when mid-air collisions were an urgent political crisis.&lt;/p&gt;
&lt;p&gt;Under the hood, an IBM 9020 tied together six or seven modified System/360 compute elements sharing memory through dedicated Storage Elements, orchestrated by a real-time control program managing hundreds of radar and display peripherals. Its standout engineering achievement was fault tolerance: when a hardware fault occurred, the Operational Error Analysis Program (OEAP) automatically diagnosed the failure, ignored transient glitches, or rewrote configuration registers on the fly to isolate a failed compute or storage element so air traffic control could continue uninterrupted. The 9020 stayed in service into the mid-1980s—with display subsystems lasting into the 1990s—proving that commercial-grade computer architectures could handle life-critical workloads when paired with disciplined redundancy and fault-recovery software.&lt;/p&gt;
&lt;h2 id="sfq-simple-stateless-stochastic-fairness"&gt;&lt;a href="https://brooker.co.za/blog/2026/02/25/sfq.html" rel="noopener noreferrer"&gt;SFQ: Simple, Stateless, Stochastic Fairness&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Marc Brooker highlights Stochastic Fairness Queuing (SFQ), an algorithm from a 1990 paper by Paul McKenney and one of his favorite compact building blocks for distributed systems. Classic fair queuing maintains a separate queue per tenant, requiring O(tenants) queues and matching scheduling overhead—impractical at cloud scale. SFQ instead allocates a fixed O(1) array of queues and hashes each tenant ID to a queue. Because two tenants that hash to the same bucket would stay stuck together if one is a &amp;ldquo;noisy neighbor,&amp;rdquo; SFQ periodically rotates the hash seed so tenants that collide in one epoch are overwhelmingly unlikely to collide in the next.&lt;/p&gt;
&lt;p&gt;Brooker then describes a modern variant combining SFQ with shuffle sharding and the power of two choices: each tenant is hashed to a small subset of queues (often just two), each incoming request is placed into whichever of those queues is shorter, and the hash seed rotates periodically. The result keeps queue count, enqueue cost, and dequeue cost at O(1) while strongly isolating noisy neighbors from well-behaved tenants—as long as noisy tenants are a small fraction of the population. The pattern works both inside a single multi-tenant server and across a fleet of backend workers.&lt;/p&gt;
&lt;h2 id="how-many-branches-can-your-cpu-predict"&gt;&lt;a href="https://lemire.me/blog/2026/03/18/how-many-branches-can-your-cpu-predict/" rel="noopener noreferrer"&gt;How many branches can your CPU predict?&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Daniel Lemire measures how many conditional branches a modern CPU&amp;rsquo;s branch predictor can &amp;ldquo;memorize.&amp;rdquo; His benchmark loops over an array of random integers and conditionally writes to a buffer only when a value is odd—so on fresh random data, the CPU mispredicts roughly 50% of the time. However, if you repeatedly loop over the &lt;em&gt;exact same&lt;/em&gt; random sequence, the branch predictor learns the pattern; increasing the sequence length reveals the capacity threshold where the predictor overflows and performance drops back to random guessing. The results: AMD Zen 5 perfectly predicts sequences of roughly 30,000 branches, Apple&amp;rsquo;s M4 handles around 10,000, and Intel&amp;rsquo;s Emerald Rapids manages only about 5,000—six times fewer than Zen 5.&lt;/p&gt;
&lt;p&gt;The practical lesson is a warning for microbenchmarking: if your test input is small enough to fit inside the CPU&amp;rsquo;s branch history tables, the processor will memorize the exact branch pattern and report numbers far faster than what you will ever see on real production traffic. Reliable benchmarks require input datasets large and varied enough to defeat branch memorization.&lt;/p&gt;
&lt;h3 id="bonus"&gt;Bonus&lt;/h3&gt;&lt;p&gt;&lt;strong&gt;Images:&lt;/strong&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!tS1T!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F958d36d9-6b19-4949-b6a6-6aa614ede834_2508x2960.jpeg" alt="Different Types of Tests"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!c4WE!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd86cd367-494f-4031-9793-f27e6142e54b_2508x3000.png" alt="How Single Sign-On (SSO) Works"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!em-f!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7e7e23e9-2a5a-4b06-a3fa-ccb6c5493f1d_2508x3000.png" alt="How LLMs Use AI Agents with Deep Research"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!kRPi!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3d3b2bb9-fc09-48d9-89b7-52bc37098af9_2360x2960.png" alt="How hackers steal passwords"&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;&lt;em&gt;This post was rewritten by Claude Code with Opus 5.5 on 27/09/2026; reviewed and edited by Codex with GPT 6.1 Sol on 09/11/2026, and re-verified by Claude Code with Sonnet 5.5 on 16/11/2026.&lt;/em&gt;&lt;/p&gt;</description></item><item><title>Newsletter #77</title><link>https://miti99.com/en/post/2026/01/24/</link><pubDate>Sat, 24 Jan 2026 00:00:00 +0700</pubDate><guid>https://miti99.com/en/post/2026/01/24/</guid><description>&lt;p&gt;&lt;em&gt;Enjoy Newsletter #77.&lt;/em&gt;&lt;/p&gt;
&lt;h2 id="stop-forwarding-errors-start-designing-them"&gt;&lt;a href="https://fast.github.io/blog/stop-forwarding-errors-start-designing-them/" rel="noopener noreferrer"&gt;Stop Forwarding Errors, Start Designing Them&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;This article argues that most Rust developers merely &amp;ldquo;forward&amp;rdquo; errors rather than truly handling them: catching an error, wrapping it thinly, and propagating it up the call stack as fast as possible. The original low-level message survives intact, but almost all higher-level context is lost. The author points out the limitations of familiar tools: &lt;code&gt;std::error::Error&lt;/code&gt; assumes a linear chain of causes via &lt;code&gt;source()&lt;/code&gt; even though real-world failures often have multiple contributing causes; stack traces are expensive, misleading in async code, and only show where an error originated rather than the logical path execution took. Meanwhile, &lt;code&gt;thiserror&lt;/code&gt; categorizes errors by origin rather than by how callers can react, &lt;code&gt;anyhow&lt;/code&gt; makes adding context optional so it is routinely skipped, and the dynamic &lt;code&gt;Provide&lt;/code&gt;/&lt;code&gt;Request&lt;/code&gt; API makes behavior hard to predict.&lt;/p&gt;
&lt;p&gt;The proposed solution is to design errors for two distinct audiences. For machines, use a flat error struct categorized by error kind (similar to Apache OpenDAL&amp;rsquo;s design) with explicit retryability status—such as permanent, temporary, or persistent—so code can decide whether to retry or abort without walking nested error chains. For humans, use tree-structured context tracking that automatically records call sites via &lt;code&gt;#[track_caller]&lt;/code&gt; at near-zero cost and enforces adding context across module boundaries through the type system (as in the &lt;code&gt;exn&lt;/code&gt; crate). The guiding question when designing errors should be: &amp;ldquo;If the system breaks at 3 a.m., what do you want the logs to say?&amp;rdquo;&lt;/p&gt;
&lt;h2 id="clock-synchronization-is-a-nightmare"&gt;&lt;a href="https://arpitbhayani.me/blogs/clock-sync-nightmare/" rel="noopener noreferrer"&gt;Clock Synchronization Is a Nightmare&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;This post explains why synchronizing time across distributed systems is far harder than intuition suggests: there simply is no &amp;ldquo;global clock.&amp;rdquo; Every computer keeps time using a quartz crystal oscillating at 32,768 Hz, but manufacturing tolerances and temperature swings cause clocks to drift; two machines started at the exact same instant can drift hundreds of milliseconds apart after a single day. Instantaneous clock skew between two machines leads to subtle bugs: in a distributed &lt;code&gt;make&lt;/code&gt; build, a compiled object file can appear newer than a freshly edited source file and fail to recompile; in banking, a withdrawal timestamped earlier than a preceding deposit can make an account appear overdrawn.&lt;/p&gt;
&lt;p&gt;Physical clock synchronization techniques include Cristian&amp;rsquo;s algorithm (estimating one-way network delay as half the round-trip time), the Berkeley algorithm (averaging time across nodes and sending relative adjustments), hierarchical NTP backed by atomic or GPS clocks (achieving millisecond accuracy over the internet), and PTP using hardware timestamping on network interface cards for sub-microsecond accuracy. Google Spanner&amp;rsquo;s TrueTime returns a bounded uncertainty interval &lt;code&gt;[earliest, latest]&lt;/code&gt; instead of a single point in time. Logical clocks include Lamport Timestamps, Vector Clocks that capture causality rather than wall-clock time, and Hybrid Logical Clocks (used in CockroachDB) that combine both. The takeaway: there is no perfect clock sync, only trade-offs between accuracy, latency, and complexity depending on your system&amp;rsquo;s needs.&lt;/p&gt;
&lt;h2 id="java---learning-roadmap"&gt;&lt;a href="https://nemorize.com/roadmaps/java" rel="noopener noreferrer"&gt;Java - Learning Roadmap&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;This is a structured Java learning roadmap on the Nemorize platform that guides learners from foundational setup through enterprise application development. It begins with prerequisites like basic Linux, Git version control, and choosing an IDE, then moves into object-oriented programming (classes, objects, inheritance, polymorphism, encapsulation, abstraction), and covers core language features including the Collections Framework, exception handling, and Streams. Advanced topics include concurrency and multithreading, JVM internals, design patterns, and Dependency Injection.&lt;/p&gt;
&lt;p&gt;The final sections focus on professional engineering skills: testing (unit, integration, contract), build tools like Maven and Gradle, working with databases, Spring Boot for enterprise applications, and principles like Clean Code, SOLID, API design, and system monitoring. Each topic follows a three-step loop: read the lesson, practice with quizzes, and track progress (requires sign-in). At the time of writing, roughly 14 of the 44 topics have full lessons available with the rest being added progressively, making it a useful reference roadmap for beginners wondering what order to learn Java in.&lt;/p&gt;
&lt;h2 id="the-ultimate-guide-to-successfully-adopting-kotlin-in-a-java-dominated-environment"&gt;&lt;a href="https://blog.jetbrains.com/kotlin/2025/12/the-ultimate-guide-to-successfully-adopting-kotlin-in-a-java-dominated-environment/" rel="noopener noreferrer"&gt;The Ultimate Guide to Successfully Adopting Kotlin in a Java-Dominated Environment&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;JetBrains&amp;rsquo; guide frames introducing Kotlin into a predominantly Java organization not as flipping a switch or rewriting everything from scratch, but as a challenge of people, timing, risk management, and trust. The roadmap spans five stages. First, get comfortable with Kotlin in test suites (using Kotest and MockK) where production isn&amp;rsquo;t at risk, answering the question, &amp;ldquo;Is this language genuinely nicer to work with?&amp;rdquo; Next, evaluate Kotlin in real production work by building a small new service or adding a Kotlin module to an existing system while avoiding the trap of &amp;ldquo;writing Java in Kotlin syntax&amp;rdquo;: use extension functions instead of static utility classes, nullable types instead of &lt;code&gt;Optional&lt;/code&gt;, and data classes instead of boilerplate POJOs.&lt;/p&gt;
&lt;p&gt;Stage three is building internal momentum through short demos, starter template repositories, and pair programming. Stage four is making the business case to management: less code to maintain, fewer bugs thanks to null safety, 100% Java interoperability so no big-bang rewrite is required, and predictable ramp-up costs. Finally, when scaling adoption across the organization, treat each system according to its lifecycle: leave sunsetting applications alone, default to Kotlin for greenfield projects, and migrate active systems incrementally with the help of IDE conversion tools, JSpecify nullability annotations, and AI-assisted refactoring.&lt;/p&gt;
&lt;h2 id="rust-vs-go"&gt;&lt;a href="https://bitfieldconsulting.com/posts/rust-vs-go" rel="noopener noreferrer"&gt;Rust vs Go&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;John Arundel compares Rust and Go across performance, simplicity, safety, features, scalability, and concurrency, summing up the trade-off in a single phrase: &amp;ldquo;Rust for high stakes, Go for low cost.&amp;rdquo; Both are compiled languages that produce small, fast native binaries and come with great built-in tooling (&lt;code&gt;gofmt&lt;/code&gt;, &lt;code&gt;rustfmt&lt;/code&gt;). Their difference lies in philosophy: Go is a small, easy-to-learn, fast-compiling garbage-collected language that prioritizes developer velocity and stability, with goroutines and channels making concurrency remarkably approachable. Rust prioritizes correctness, peak performance, and low-level control, using its borrow checker to eliminate memory bugs at compile time without a garbage collector, at the cost of a steeper learning curve.&lt;/p&gt;
&lt;p&gt;For error handling, Go uses explicit &lt;code&gt;if err != nil&lt;/code&gt; checks, while Rust uses &lt;code&gt;Option&lt;/code&gt;/&lt;code&gt;Result&lt;/code&gt; types paired with the &lt;code&gt;?&lt;/code&gt; operator. Go shines when fast onboarding, rapid prototyping, and cost-effective delivery matter most—such as web services, microservices, infrastructure tooling, and business automation. Rust shines when reliability, predictable latency, and resource efficiency are mission-critical—such as automotive, aerospace, medical devices, and industrial systems. He views the two languages as complementary tools rather than rivals, and to the question &amp;ldquo;Should I learn Rust or Go?&amp;rdquo;, his answer is: learn both.&lt;/p&gt;
&lt;h2 id="the-most-popular-go-dependency-is"&gt;&lt;a href="https://blog.thibaut-rousseau.com/blog/the-most-popular-go-dependency-is/" rel="noopener noreferrer"&gt;The most popular Go dependency is…&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Thibaut Rousseau sets out to find the most depended-upon library in the Go ecosystem by tapping into Go&amp;rsquo;s centralized module proxy infrastructure. Instead of cloning individual repositories, he downloaded the full module index from &lt;code&gt;index.golang.org&lt;/code&gt; (listing every module version published since the Go proxy launched in 2019), fetched each module&amp;rsquo;s &lt;code&gt;go.mod&lt;/code&gt; file via &lt;code&gt;proxy.golang.org&lt;/code&gt; to extract its dependencies, and loaded everything into a Neo4j graph database. The resulting graph contains roughly 40 million nodes and 400 million relationships, averaging about 10 direct dependencies per module version node in the dataset.&lt;/p&gt;
&lt;p&gt;Topping the leaderboard is &lt;code&gt;testify&lt;/code&gt; with 259,237 dependent modules, far ahead of the runners-up: &lt;code&gt;google/uuid&lt;/code&gt; (104,877), &lt;code&gt;golang.org/x/crypto&lt;/code&gt; (100,633), &lt;code&gt;grpc&lt;/code&gt; (97,228), &lt;code&gt;cobra&lt;/code&gt; (93,062), &lt;code&gt;pkg/errors&lt;/code&gt; (92,491), &lt;code&gt;golang.org/x/net&lt;/code&gt;, &lt;code&gt;protobuf&lt;/code&gt;, &lt;code&gt;logrus&lt;/code&gt;, and &lt;code&gt;viper&lt;/code&gt;. Testing and utility libraries dominate the list, alongside Go&amp;rsquo;s extended standard library packages (&lt;code&gt;golang.org/x/&lt;/code&gt;). Notably, even though &lt;code&gt;pkg/errors&lt;/code&gt; has been deprecated and archived for years, version &lt;code&gt;v0.9.1&lt;/code&gt; alone still had 16,001 transitive dependents in 2025. The post also demonstrates how to query Neo4j with Cypher to trace direct and transitive dependencies, showing how naturally graph databases fit dependency analysis. He shares the dataset dump via BitTorrent so the community can run their own queries and plans to enrich it with metadata like GitHub stars.&lt;/p&gt;
&lt;h2 id="gosum-is-not-a-lockfile"&gt;&lt;a href="https://words.filippo.io/gosum/" rel="noopener noreferrer"&gt;go.sum Is Not a Lockfile&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Filippo Valsorda clears up a widespread misconception: many developers treat &lt;code&gt;go.sum&lt;/code&gt; as a lockfile analogous to Node&amp;rsquo;s &lt;code&gt;package-lock.json&lt;/code&gt; or Rust&amp;rsquo;s &lt;code&gt;Cargo.lock&lt;/code&gt;, when in reality &lt;code&gt;go.sum&lt;/code&gt; is merely a local cache of the Go Checksum Database—a lookup table mapping module versions to their cryptographic hashes. Versions listed in &lt;code&gt;go.sum&lt;/code&gt; may or may not actually be used in a build, and the file plays zero role in version resolution; in the original Go modules design, it wasn&amp;rsquo;t even enabled by default because it has no effect on build selection. The sole purpose of &lt;code&gt;go.sum&lt;/code&gt; is security: the Checksum Database ensures the entire ecosystem receives identical bytes for a given module version, and &lt;code&gt;go.sum&lt;/code&gt; makes that guarantee work locally and offline.&lt;/p&gt;
&lt;p&gt;The file you actually want to look at is &lt;code&gt;go.mod&lt;/code&gt;, which serves as both manifest and lockfile. Since Go 1.17 (August 2021), &lt;code&gt;go.mod&lt;/code&gt; records every transitive dependency needed to build the main module and run its tests, pinned to exact versions in a single human-readable file. Combined with Minimal Version Selection (MVS), this lets Go avoid diamond dependency conflicts, prevents accidentally relying on features from newer transitive versions than declared, and never silently pulls the latest (potentially compromised) version of a package when adding a new dependency—all while making dependency resolution so fast that nobody even notices it. To analyze a Go dependency graph, parse &lt;code&gt;go.mod&lt;/code&gt; using &lt;code&gt;golang.org/x/mod/modfile&lt;/code&gt; or &lt;code&gt;go mod edit -json&lt;/code&gt;, and never parse &lt;code&gt;go.sum&lt;/code&gt;.&lt;/p&gt;
&lt;h2 id="forgejo-actions-notification-development"&gt;&lt;a href="https://chris-besch.com/articles/forgejo_actions_notification/" rel="noopener noreferrer"&gt;Forgejo Actions Notification Development&lt;/a&gt;&lt;/h2&gt;&lt;p&gt;Christopher Besch shares his experience contributing to Forgejo, a self-hosted software forge similar to GitHub and GitLab. He needed email and webhook notifications when CI workflows failed—a feature Forgejo lacked—so he built it himself. The post explains how Go projects organize code into modules (via &lt;code&gt;go.mod&lt;/code&gt;) and packages without a central registry like npm, then walks through Forgejo&amp;rsquo;s layered architecture consisting of &lt;code&gt;/routers&lt;/code&gt; (APIs and web UI), &lt;code&gt;/services&lt;/code&gt; (business logic), &lt;code&gt;/models&lt;/code&gt; (database access), and &lt;code&gt;/modules&lt;/code&gt; (self-contained utilities), where lower layers are forbidden from importing higher layers. To avoid circular dependencies, Forgejo uses a publish–subscribe pattern in &lt;code&gt;forgejo.org/services/notify&lt;/code&gt;: packages publish events to a topic, and interested packages subscribe to it.&lt;/p&gt;
&lt;p&gt;He split the feature across four pull requests to keep reviews manageable: PR #7510 moved code from &lt;code&gt;/models/actions&lt;/code&gt; to &lt;code&gt;/services/actions&lt;/code&gt; to respect architectural boundaries, PR #7491 added an &lt;code&gt;ActionRunNowDone&lt;/code&gt; hook to the pub-sub notifier, PR #7509 subscribed the mailer service to send emails when workflows fail or recover, and PR #7508 added webhook delivery. He also recounts having to rename a struct due to a merge conflict with a parallel PR. The final section walks through setting up a local development environment: compiling Forgejo with debug symbols, running the Forgejo Runner, spinning up a MailDev SMTP server in Docker, writing a tiny Node.js webhook receiver script, and running unit tests, integration tests, and the Delve debugger. His takeaway: thorough tests are the key to earning maintainers&amp;rsquo; trust in your contributions.&lt;/p&gt;
&lt;h3 id="bonus"&gt;Bonus&lt;/h3&gt;&lt;p&gt;&lt;strong&gt;Images:&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;&lt;img src="https://substackcdn.com/image/fetch/w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3ec94ed4-46b2-40bd-8d28-a25fdde639c8_2250x2624.png" alt="Message Brokers 101"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!hXRV!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe99863c7-7ab7-4f61-9961-af7e4c6ee64f.png" alt="Cloud Load Balancer Cheat Sheet"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!XvtW!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1ded3402-4eb5-4818-861f-7661f1cefe82.tif" alt="How CQRS Works?"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!SUjA!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd91a5538-a69e-468c-bf83-058bb78753ca_2360x2770.png" alt="How does Docker Work?"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!4qBZ!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff0945414-9cea-4e64-8dd4-3abcc404f73e_2360x2960.png" alt="Containerization Explained: From Build to Runtime"&gt;
&lt;img src="https://substackcdn.com/image/fetch/$s_!n8BK!,w_1100,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa51a75cc-09a9-4fed-af8a-c32a64f8ab60_2250x2624.png" alt="Must-Know Message Broker Patterns"&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;&lt;em&gt;This post was rewritten by Claude Code with Opus 5.5 on 27/09/2026; reviewed and edited by Codex with GPT 6.1 Sol on 09/11/2026, and re-verified by Claude Code with Sonnet 5.5 on 16/11/2026.&lt;/em&gt;&lt;/p&gt;</description></item></channel></rss>