目录
[1. Team Division of Labor and Collaboration](#1. Team Division of Labor and Collaboration)
[2. Development Process & Tooling](#2. Development Process & Tooling)
[2.1 Verified Improvements](#2.1 Verified Improvements)
[2.2 Beta-Phase Landing Results](#2.2 Beta-Phase Landing Results)
[3. Product Functionality & Implementation](#3. Product Functionality & Implementation)
[3.1 Verified Improvements](#3.1 Verified Improvements)
[3.2 Beta-Phase Landing Results](#3.2 Beta-Phase Landing Results)
[4. Quality Assurance: JMeter at the Core](#4. Quality Assurance: JMeter at the Core)
[4.1 JMeter](#4.1 JMeter)
[4.2 Beta-Phase QA Figures](#4.2 Beta-Phase QA Figures)
[5. Blogging & Time Management](#5. Blogging & Time Management)
[5.1 Verified Improvements](#5.1 Verified Improvements)
[5.2 Beta-Phase Landing Results](#5.2 Beta-Phase Landing Results)
[6. Project-Wide Summary](#6. Project-Wide Summary)
|------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| Course | EE308FZ --- Software Engineering |
| Class Link | 2501_MU_SE_FZU社区-CSDN社区云 |
| Requirements | Teamwork---beta Spring-CSDN社区 |
| Team Name | FZU Meteorological Bureau |
| Objective | Address the problems left over from the Alpha phase |
| Git Repositories | https://github.com/wogua1517/software_engineering_front.git |
1. Team Division of Labor and Collaboration
Based on the review of the Alpha stage, our team fully implemented process improvements, technical specifications and built-in quality practices in the Beta stage. After four weeks of high-intensity iterations, all original problems have been solved in a closed loop, core functions have been stably launched, and system performance and user experience have been significantly improved. This report uses a "problem-act-verify" structure to fully present the improvements in the Beta phase, demonstr
2. Development Process & Tooling
2.1 Verified Improvements
- Tech-spike day: One-day group spike before sprint; 23 Vue3 + Spring-Boot multi-module traps captured in internal wiki.
- Git workflow hardened: main / dev / feature triad + commitlint template; merge conflicts 8 → 0.
- Shift-left testing: Test-case review before dev-start; defect detection moves 3 days earlier, fix cost −40 %.
2.2 Beta-Phase Landing Results
- New-member on-boarding 1 d → 0.5 d;
- Code-review coverage 100 %, Sonar blocker issues 0;
- Auto-test pass rate steady ≥ 97 %.
3. Product Functionality & Implementation
3.1 Verified Improvements
- 3rd-party resilience: Weather API + fallback cache + circuit-breaker; users notice no outage, SLA 99.9 %.
- Experience metrics quantified: Lighthouse first paint 1.4 s, interaction delay 120 ms, 55 % improvement vs. Alpha.
- Edge scenarios covered: Empty data, network loss, location failure---12 unified Toast patterns, zero crashes recorded.
3.2 Beta-Phase Landing Results
- P0 / P1 legacy defects 0;
- Air-quality early-warning & volunteer-hours certification added, PO acceptance passed first time;
- Mobile UI walk-through on 6 mainstream devices, issues 0.
4. Quality Assurance: JMeter at the Core
4.1 JMeter
We standardized on Apache JMeter for both load and functional testing:
- Open-source & extensible---custom weather-api sampler plug-in written in 2 h;
- GUI + CLI---engineers script in GUI, CI calls non-GUI mode, results auto-export to InfluxDB + Grafana;
- Distributed load---one controller + three slaves simulates 5 k concurrent users on a single 4-core note-book.
4.2 Beta-Phase QA Figures
- 65 test scripts, 820 assertions, 0 false positives;
- 95-th percentile response time 380 ms (target ≤ 500 ms);
- Peak throughput 1 250 req/s @ CPU < 70 %, memory < 65 %;
- Memory-leak 4-h soak test---heap growth < 2 %, no GC pause > 200 ms;
- JMeter + Jenkins pipeline gates merge: performance regression > 5 % = build fail.
5. Blogging & Time Management
5.1 Verified Improvements
- Log-as-code: PR template enforces link to "daily sprint log"; no post-sprint catch-up accepted.
- Estimate double-lock: Dev estimate → tech-lead review; >20 % delta triggers retrospective.
- Doc-sync gate: Function merged only after interface, test, deploy docs updated; no doc, no merge.
5.2 Beta-Phase Landing Results
- Blog real-time update rate 100 %, zero last-minute rush;
- Task actual vs. estimate median deviation 8 %;
- Doc / code version mismatch 0.
6. Project-Wide Summary
6.1 Technical summary
The beta phase is delivered data-driven: all alpha issues are closed loop with quantifiable metrics; JMeter implements 5k concurrent stress tests, 95% response times ≤380ms, and embeds CI as performance gate control. The development process is fully strengthened - Git conflicts are zeroed, newcomer onboarding time is halved, and automated test coverage is 100%. The product functions not only fix defects, but also update high-value modules such as air warning and volunteer duration certification, and have zero problems with UI acceptance on all platforms, with an SLA of 99.9%.
6.2 Team perception
We have moved from "doing and making mistakes" to "doing and measuring". Role boundaries are clear but flexibility is retained ("Owner + 20% flex"), interface contracts ensure zero friction for collaboration, and physical + digital signage leaves risks unhidden. What's more, documentation, logs, and estimates are no longer remediated after the fact, but are reviewed together with code. This delivers not just a system, but a set of reusable, trustworthy, and truly production-oriented engineering habits.