10 Things Every Developer Should Know About Non Sunt Atque Aut.

Listicles get an undeserved reputation in technical writing. A well-researched list with genuinely substantive entries is one of the most scannable and practically memorable formats available for communicating technical knowledge. Every item below is backed with enough depth to be immediately actionable on your next project or code review.

Bookmark this page — it is the kind of reference you will want to return to during architecture discussions and pull request reviews.

The Essential List

1. Start With the Problem Statement, Not the Solution

The single biggest mistake developers make when approaching Non Sunt Atque Aut. is reaching for a familiar tool or pattern before fully and precisely understanding the problem they are actually trying to solve. Invest twenty percent more time defining the problem with specificity and you will consistently spend fifty percent less time debugging an inadequate solution later in the project.

2. Document the Trade-Offs You Seriously Considered

Every meaningful technical decision is fundamentally a trade-off between competing priorities and constraints. There is no universal silver bullet in software engineering. Document the specific alternatives you evaluated and the concrete reasons you made your final choice — this context becomes invaluable six months or a year later when a new team member asks why the system works the way it does.

3. Design Explicitly and Deliberately for Failure

Assume with complete confidence that every external dependency your system relies upon will fail at some point, under some combination of conditions — because statistically it absolutely will. Design your system so that individual component failures degrade functionality gracefully rather than cascading into complete system outages. Circuit breakers, sensible fallback responses, and retry logic with exponential backoff are the foundational tools for achieving this resilience.

4. Instrument Everything From the Very Beginning

You cannot effectively debug, optimise, or make evidence-based architectural decisions about what you cannot observe. Instrument your application with structured logging, performance metrics, and distributed tracing from the very first day of the project. The cost of adding comprehensive observability retroactively is consistently an order of magnitude higher than building it in as a first-class concern from the start.

5. Automate Every Repetitive Manual Step in Your Workflow

Every manual step in your development or deployment workflow is simultaneously a potential source of human error and a constant drain on your team's focus and collective morale. Automate your CI/CD pipelines, database migration execution, environment provisioning scripts, and repetitive code generation tasks. The time you invest in solid automation tooling compounds over the entire lifetime of the project in ways that are genuinely difficult to overstate.

6. Keep All Dependencies Current and Regularly Audited

Falling significantly behind on dependency updates is a form of technical debt that is deceptively easy to ignore month by month and surprisingly painful to resolve all at once when security vulnerabilities are discovered. Set up automated dependency update pull requests using tools like Dependabot or Renovate, and allocate a regular dedicated slot in your sprint for reviewing and merging them before they accumulate into an overwhelming backlog.

7. Write Documentation Directly Alongside the Code That Justifies It

Documentation written weeks or months after the fact is almost always incomplete, partially inaccurate, and very quickly becomes stale as the codebase continues evolving without it. Write inline comments, Architecture Decision Records, and README updates as integral components of the same pull request that introduces the change they document. Make this a hard and enforced requirement in your team's definition of done.

8. Use Code Review Primarily as a Knowledge-Sharing Exercise

The primary and most enduring value of systematic code review is knowledge transfer and shared understanding across the team, not defect detection. Use reviews as deliberate opportunities to spread important business and technical context, highlight reusable patterns worth deliberately propagating across the codebase, and provide meaningful mentorship and growth opportunities to less experienced teammates. Reviews should be genuine learning conversations, not bureaucratic approval ceremonies.

9. Respect the Principle of Least Surprise at Every Layer

Every public interface, API endpoint, module boundary, and utility function in your system should behave precisely as a developer familiar with your established conventions would naturally and confidently expect. Surprising behaviour — read-only method names that silently mutate state, endpoints that return structurally different response shapes based on undocumented hidden conditions — erodes team trust and systematically leads to defensive coding patterns that create their own cascade of downstream problems.

10. Ship Frequently in Small, Safe, Incremental Changes

Large, infrequent releases carry enormous inherent risk and deliver slow, delayed feedback about whether the changes are working as intended for real users. Small, frequent releases are demonstrably lower risk in practice and provide the fast feedback loops that allow teams to course-correct efficiently before problems compound. Trunk-based development combined with feature flags and a mature continuous deployment pipeline makes this approach entirely practical even for large distributed teams working on complex, long-running features simultaneously.

Quick Reference Checklist

• Problem fully and precisely defined before any solution approach was chosen
• Trade-offs explicitly documented in an Architecture Decision Record for future reference
• Failure modes identified, tested under realistic conditions, and handled gracefully
• Structured logging, performance metrics, and distributed tracing fully in place
• Entire deployment pipeline automated with zero manual intervention steps remaining
• All dependencies enrolled in an automated regular update schedule
• Documentation merged as a required part of every code-changing pull request
• Code review conducted deliberately as a learning and knowledge-sharing exercise
• All public interfaces consistent, predictable, and unsurprising by conscious design
• New features shipped behind feature flags for safe, controlled gradual rollout

Final Thoughts

These ten principles apply consistently and reliably across different technology stacks, team compositions, team sizes, and project types. They are not rigid rules to follow blindly in every situation but deeply useful heuristics to genuinely internalise over time through practice and reflection. The more thoroughly you understand the underlying reasoning behind each one, the better engineering judgement you will bring to the real-world situations that never quite fit neatly into any prepared checklist.

Which of these resonates most strongly with your current engineering context? Or which do you find most genuinely difficult to apply consistently in practice given your team's specific constraints? Share your honest perspective in the comments section below.