What Makes a Code Hero? Lessons from a Decade of Huawei Software Engineering

导读：练好基本功，练好内功，才是真的代码英雄。

One night the author was asked to write a “big‑shot software” article; after a humorous exchange with his wife, he decided to recount his own experiences. Over the past six months, appointments such as Software Chief Engineer and Software Expert, as well as a company directive to improve software engineering capability, have energized him.

Since joining Huawei in 2006, he has witnessed the full lifecycle of 3G controllers and, at age 35, joined the 5G development team. He has held many roles—from ordinary developer to TL, MDE, MDEL, SDM, Committer, and Software Expert—yet attributes his longevity to solid software fundamentals, meticulous work attitude, and relentless technical curiosity.

What Does Good Code Look Like?

Winning a “Gold Code” award was not about flashy tricks but about simplicity, clear logic, single‑purpose functions, and sensible data‑structure design, following the company’s C/C++ coding standards. Simple, logical code is easier to read, maintain, and adapt without introducing bugs.

Simple does not mean thoughtless. The author describes a three‑week performance‑optimization effort where careful hotspot analysis led to a small backup‑mechanism tweak that yielded significant performance gains.

Another case involved a CPU‑load increase after a version upgrade; the culprit was a direct DB‑kernel read. A caching solution would have avoided the performance penalty, illustrating the impact of thoughtful design.

He stresses that software must be extensible and adaptable to ever‑changing customer demands, requiring continuous learning, broad knowledge, and foresight into future technology trends.

Mastering the Basics

Writing good code consistently demands a deep understanding of technical principles and business logic, grounded in solid programming fundamentals such as data structures, algorithms, and compilation theory.

At an architecture conference, topics like AI, blockchain, and IoT were hot, yet the most popular sessions covered foundational software and architecture design, reinforcing the value of strong basics.

Good habits—keyboard shortcuts, effective comments, naming conventions, and coding style—also elevate programming cultivation. Small optimizations, like replacing dynamic memory allocation with a static pool, once yielded a 30 CAPS performance boost.

The Tragedy of a Single Line of Code

Careless copy‑paste and rushed development can lead to severe incidents. The author recounts two cases where accidental deletion of a single line caused major KPI regressions, prompting reflection on code‑review processes and the role of MDEs and Committers.

He explains that Committer responsibilities act as a cultural shift, encouraging engineers to improve software quality, though the ultimate goal is proactive design and debt reduction.

With automated testing and change‑protection mechanisms in place, such tragedies become avoidable.

Is the “Change Protection Wall” Reliable?

For veteran engineers, any change can feel like a battle; performance regressions are unacceptable. The author describes expanding test cases from hundreds to tens of thousands, covering scenarios, signal validity, failure judgments, and resource leaks, creating a robust protection wall for 3G.

He also shares an anecdote of deliberately introducing a bug to test the effectiveness of test cases, which succeeded, highlighting the importance of thorough validation.

Maintaining Curiosity for Emerging Technologies

Reflecting on a 2016 cloud‑migration project, the author realized his prior focus on embedded software limited his perspective. By studying distributed systems and leveraging cloud solutions from industry leaders, he delivered a satisfactory design.

This experience reinforced the need for programmers to stay curious about new technologies, especially in the demanding 5G era, where network security, ultra‑low latency, and massive connectivity raise software reliability and performance challenges.

Huawei’s five‑year plan to elevate software engineering capability calls for unified coding standards, clean architecture, and persistent dedication. The author concludes that while few become legends like Linus Torvalds, any programmer can contribute valuable code to the world’s communication history.