Maintainable

Joel Oliveira, engineering manager at ezCater with decades in software, focuses on maintainability and practical migrations. He talks about predictability in codebases, why test coverage can mislead, replacing a leaky image service with ImageProxy, finding a GraphQL seam for incremental migration, and using feature flags and small steps to safely modernize systems.

Lucas Roesler, Managing Partner and CTO at Contiamo, is a data engineering and DevOps leader focused on observability and modernizing legacy systems. He talks about designing for fast feedback through logs, metrics, and tracing. He tells a concrete decade-old Postgres modernization story. He explains treating SQL and infrastructure as code and running local observability to reduce guesswork.

Brittany Ellich, a Senior Software Engineer at GitHub, talks about the importance of maintaining existing software rather than opting for risky rewrites. She emphasizes that well-maintained software is readable and manageable for future developers. Brittany redefines technical debt, arguing it often includes bugs and reliability issues. The conversation highlights how AI, like coding agents, can assist in tackling small maintenance tasks incrementally while leaving complex problem-solving to humans. Her insights challenge conventional views on software evolution and ethical AI use.

Kent Beck, a pioneer in agile methods and author of Tidy First?, dives into the critical but often overlooked concept of optionality in software development. He discusses the tension between current features and maintaining future flexibility, highlighting how developers must balance these while navigating cost shifts from cheap labor to expensive compute. Beck advises on making hard changes easier, avoiding splitting teams, and emphasizes the ethical duty to enhance code clarity to benefit the entire team. His insights challenge traditional views on software maintainability.

Don MacKinnon, a software engineer and founder of Searchcraft, discusses the importance of simplicity in software design. He shares a story about how an unnecessary abstraction in a Node.js API complicated development. Don explains how his team’s structure evolved from consulting pain points and emphasizes the value of clear project structure and documentation. He also advises on evaluating third-party libraries and uses real-world examples to illustrate the transition from dynamic languages to TypeScript and Rust for improved maintainability.

In this conversation, Chris Zetter, a software engineer and author focused on architecture and maintainability, shares insights about building relational databases. He emphasizes that effective software should enhance value and adaptability, often prioritizing solid documentation over code comments. Chris challenges the typical notions of technical debt, advocating for transparency about its causes. He also discusses the benefits of pairing in coding, especially for junior developers, and the value of choosing stable technology to improve reliability.

Maintaining consistency across a sprawling codebase is one of the hardest challenges in software engineering. Denis Rechkunov, a Principal Software Engineer at Elastic, joins Robby to share how his team turned consistency into a cultural practice rather than a technical checklist. From managing open source projects with hundreds of contributors to experimenting safely with new patterns, Denis believes maintainability begins with shared ownership, not just clean code.He explains how Elastic introduced automation and linters to improve cohesion without discouraging creativity. Instead of enforcing perfection across the entire system, Denis’ team scopes their changes to manageable areas and rewards steady progress over sweeping rewrites. Their annual “On Week” tradition gives engineers space to fix what frustrates them most, showing how small, focused bursts of work can produce big leaps in stability and morale.The conversation also explores the human side of maintainability. Denis recalls early lessons about unclear expectations, the importance of documenting decisions in public pull requests, and how open feedback loops build trust across remote teams. Whether it’s stabilizing a flaky CI pipeline or mentoring new engineers, Denis argues that technical excellence thrives when consistency becomes a habit shared by everyone.Episode Highlights[00:01:02] Defining Well-Maintained SoftwareDenis identifies consistency, documentation, testability, and agility as the key ingredients of maintainable systems.[00:02:22] Balancing Standards and AutonomyHow automation and linters help preserve code cohesion while minimizing interpersonal friction.[00:04:08] Experimenting SafelyElastic scopes new patterns to low-risk modules before broader adoption, avoiding mass rewrites.[00:07:19] Incremental CleanupLinters only apply to changed files, helping the team fix issues gradually without overwhelming contributors.[00:08:02] Maintainability as a People ProblemDenis highlights that sustainable systems depend more on culture and mentorship than on architecture.[00:10:13] Lessons from MiscommunicationAn early experience showed the cost of undocumented conventions and unclear onboarding.[00:17:09] Making Space for Technical DebtElastic’s engineers dedicate part of each sprint and an annual “On Week” to tackle maintenance work.[00:23:05] Restoring CI ReliabilityDenis shares how the team revived a pipeline with only a 10% success rate by categorizing failures and focusing on data.[00:32:00] Practicing Software ArchaeologyHe stresses the value of documenting discussions in pull requests to avoid historical guesswork later.[00:36:09] Feedback and TrustOpen communication, humility, and mutual feedback loops form the backbone of a maintainable culture.[00:51:00] Embracing Chaos in Open SourceDenis encourages teams to accept a degree of entropy and focus their efforts on user-facing stability.[01:00:00] Security and PrivacyWhy maintainability, trust, and privacy are inseparable pillars of long-term sustainability.[01:01:06] Where to StartInstead of rewriting code, start by cultivating maintainability as a shared value across the team.Resources MentionedElasticgolangci-lintAppSignalThe Caves of Steel by Isaac Asimov — Denis’ recommendation inspired Robby to finally pick up a copy and start reading it himself.Denis’s Blog – rdner.deDenis on GitHubDenis on MastodonDenis on LinkedInThanks to Our Sponsor!Turn hours of debugging into just minutes! AppSignal is a performance monitoring and error-tracking tool designed for Ruby, Elixir, Python, Node.js, Javascript, and other frameworks.It offers six powerful features with one simple interface, providing developers with real-time insights into the performance and health of web applications.Keep your coding cool and error-free, one line at a time! Use the code maintainable to get a 10% discount for your first year. Check them out! Subscribe to Maintainable on:Apple PodcastsSpotifyOr search "Maintainable" wherever you stream your podcasts.Keep up to date with the Maintainable Podcast by joining the newsletter.

Episode NotesThe discussion moves into how standards evolve beyond tools, the trade-offs of monocultures vs. consensus-driven teams, and why ownership matters when the original authors move on. Nathan also unpacks the cost of neglect, describing defects as anything that slows developers down—not just issues that impact end users.Later in the conversation, Nathan recounts a migration from a React SPA to Turbo and Stimulus that removed barriers between designers and developers. He highlights how keeping all problems on the radar together prevents teams from falling into local optima. The episode closes with reflections on TestBench, blind spots in testing, continuous improvement in remote teams, and advice for developers who feel stuck raising maintenance concerns.Episode Highlights[00:01:07] Defining Well-Maintained Software: Nathan shares his three key markers—up-to-date dependencies, adherence to team standards, and fixing defects immediately.[00:02:53] From Tools to Tacit Knowledge: Why norms start with tool-enforced rules like RuboCop but evolve into cultural agreements within teams.[00:04:49] Speed vs. Durability: Teams built on monoculture move quickly early on, but diverse, consensus-driven cultures go farther.[00:11:11] Owning the Architecture: When original developers leave, new teams must take responsibility for architecture rather than defer decisions.[00:13:37] The Cost of Neglect: Dependencies, drifting standards, and defects interact in compounding ways. Nathan reframes defects as “anything that impedes developer effectiveness.”[00:17:46] React → Turbo + Stimulus Migration: A costly SPA and siloed design team gave way to a simpler approach that reduced rework and empowered designers to contribute directly.[00:22:44] Avoiding Local Optima: Tackling problems in isolation creates dead ends—addressing them holistically opens real paths forward.[00:24:32] Who We Seek Validation From: Developer identities often align with whose approval they value—shaping front-end vs. back-end divides.[00:27:34] Comfort vs. Maintenance Burden: Silos built for comfort create tomorrow’s maintenance problems.[00:33:45] Relentless Improvement in Remote Teams: Start as an ensemble, evolve into autonomous work cells, and use work logs to sustain consensus.[00:38:33] What’s Missing from Remote Work: Nathan reflects on lost “hallway conversations” and the challenge of building social glue remotely.[00:40:50] The Story Behind TestBench: Dissatisfaction with existing frameworks and a desire for simplicity led to TestBench’s creation.[00:47:38] Testing Blind Spots: The biggest blind spot is equating testing with automation—interactive testing and intelligible output remain essential.[00:50:35] Advice for Stuck Engineers: Nathan encourages developers to study quality traditions, connect with peers, and embrace continuous improvement.[00:53:16] Book Recommendations: Deming’s Out of the Crisis and The New Economics, Toyota’s product development work, and Rawls’ A Theory of Justice.Tools & Resources MentionedBrightworks Digital – Nathan’s current company, where he serves as Principal.Nathan Ladd on LinkedIn – Connect with Nathan and follow his work.TestBench – A Ruby testing framework co-created by Nathan.Turbo – Hotwire framework for building modern, fast applications without heavy JavaScript.Stimulus – A modest JavaScript framework for enhancing HTML with small, reusable controllers.RSpec – A popular Ruby testing tool for behavior-driven development.Minitest – A simple and fast Ruby testing framework.RuboCop – A Ruby static code analyzer and formatter.Lessons Learned in Software Testing – Classic book on testing by Cem Kaner, James Bach, and Bret Pettichord.Out of the Crisis – W. Edwards Deming’s influential work on quality and systems thinking.The New Economics – Deming’s follow-up book on continuous improvement.A Theory of Justice – John Rawls’ seminal work on moral and political philosophy.The Toyota Product Development System – Insights into Toyota’s continuous improvement and development practices.Thanks to Our Sponsor!Turn hours of debugging into just minutes! AppSignal is a performance monitoring and error-tracking tool designed for Ruby, Elixir, Python, Node.js, Javascript, and other frameworks.It offers six powerful features with one simple interface, providing developers with real-time insights into the performance and health of web applications.Keep your coding cool and error-free, one line at a time! Use the code maintainable to get a 10% discount for your first year. Check them out! Subscribe to Maintainable on:Apple PodcastsSpotifyOr search "Maintainable" wherever you stream your podcasts.Keep up to date with the Maintainable Podcast by joining the newsletter.

Taylor Otwell, the creator of Laravel and CEO of Laravel LLC, shares his insights from 14 years of building one of the most popular web frameworks. He emphasizes simplicity and adaptability in software design, advocating for maintainability to prevent overcomplexity. Taylor recounts the early days of Laravel and the unexpected traction that transformed it into a full-time business. He discusses balancing community needs with his own preferences while navigating leadership challenges and community feedback, illustrating the evolving nature of both Laravel and his role within it.

Robby is joined by Sara Jackson, Senior Developer at thoughtbot, to explore the practical ways teams can foster resilience—not just in their infrastructure, but in their everyday habits. They talk about why documentation is more than a chore, how to build trust in test suites, and how Chaos Engineering at the application layer can help make the case for long-term investment in maintainability.Sara shares why she advocates for writing documentation on day one, how “WET” test practices have helped her avoid brittle test suites, and why she sees ports as a powerful alternative to full rewrites. They also dive into why so many teams overlook failure scenarios that matter deeply to end users—and how being proactive about those situations can shape better products and stronger teams.Episode Highlights[00:01:28] What Well-Maintained Software Looks Like: Sara champions documentation that’s trusted, updated, and valued by the team.[00:07:23] Invisible Work and Team Culture: Robby and Sara discuss how small documentation improvements often go unrecognized—and why leadership buy-in matters.[00:10:34] Why Documentation Should Start on Day One: Sara offers a “hot take” about writing things down early to reduce cognitive load.[00:16:00] What Chaos Engineering Really Is: Sara explains the scientific roots of the practice and its DevOps origins.[00:20:00] Application-Layer Chaos Engineering: How fault injection can reveal blind spots in the user experience.[00:24:36] Observability First: Why you need the right visibility before meaningful chaos experiments can begin.[00:28:32] Pitching Resilience to Stakeholders: Robby and Sara explore how chaos experiments can justify broader investments in system quality.[00:33:24] WET Tests vs. DRY Tests: Sara explains why test clarity and context matter more than clever abstractions.[00:40:43] Working on Client Refactors: How Sara approaches improving test coverage before diving into major changes.[00:42:11] Rewrite vs. Refactor vs. Port: Sara introduces “porting” as a more intentional middle path for teams looking to evolve their systems.[00:50:45] Delete More Code: Why letting go of unused features can create forward momentum.[00:51:13] Recommended Reading: Being Wrong by Kathryn Schulz.Resources & LinksSara on MastodonthoughtbotRubyConf 2024 Talk – Chaos Engineering on the Death StarBook: Being Wrong by Kathryn SchulzFlu Shot on GitHubChaosRB on GitHubSemian from Shopify — a chaos engineering toolkit for RubyThanks to Our Sponsor!Turn hours of debugging into just minutes! AppSignal is a performance monitoring and error-tracking tool designed for Ruby, Elixir, Python, Node.js, Javascript, and other frameworks.It offers six powerful features with one simple interface, providing developers with real-time insights into the performance and health of web applications.Keep your coding cool and error-free, one line at a time! Use the code maintainable to get a 10% discount for your first year. Check them out!  Subscribe to Maintainable on:Apple PodcastsSpotifyOr search "Maintainable" wherever you stream your podcasts.Keep up to date with the Maintainable Podcast by joining the newsletter.