- 評估software engineering team performance嘅心法
- Focus on output，而唔係input
- line of code 同加左幾多班唔係一個好嘅指標
- Focus on team, not individual
- 不良嘅指標
- Line of codes
- 速度
- 每條team，每個task都唔同，唔係客觀嘅標準
- 會減低同其他團隊合作嘅動機，因為會減慢自己嘅速度
- 利用率
- 唔係話每個人D task都排滿就係好
- 排滿task會少左彈性
- 正面例子
- Product delivery
- 質量
- 簡單黎講就係deliver 出黎嘅價值
- 好難有一個廣泛嘅定義，每個地方嘅情況都唔同
- Other learning
- What is high performer?
- Spent less time on rework and unplanned work
History
Chaos engineering 源自Netflix，由於cloud architecture太過複雜，好難debug，因此用「故障注入」去揾出系統嘅弱點。
Steps
- 預測系統係出現故障時嘅行為
- 針對故障去設計測試
- 制定rollback plan
- 評估故障嘅影響，留意有冇出現非預期嘅副作用
- 修複找到嘅潛在問題
常見對分布式系統做出嘅錯誤假設
- 網絡是可靠
- 延遲係零
- Bandwidth係無限
- 網絡係安全
- Topology係不變
- 有管理員handle問題
- 傳輸成本係零
- 網絡係homogeneous
測試嘅執行順序
第一個known係注意到，第二個known係了解
- Known known
- Known unknown
- Unknown known
- Unknown unknown
- IBM報告表示每一元對可用性的投資都會帶出10~100元的回收
- 6 steps
- 建一個原型或產品來測試
- 起草測試計畫–兩個重要問題: 要測試什麼, 如何度量結果
- 召集測試參與者
- 執行測試
- 到企業安排現場有人主持與指引的測試，遠端在家平時樣子的測試
- 將測試結果做紀錄與文件化
- 修正問題並改善性能
16 Jan 2019
Can read the origin article
以下無需重讀
- Minimum require for a PR to be merged
- At least one ppl review
- At least one code owner review
- Don’t help other fix bug
- Cannot help other grow
- Discourage others
- What to check
- Code style
- Readability
- Corner case (need senior ppl diswho familiar the business logic)
- Error handling
- Test case (need to review test case too!)
- Mini architecture (repeat code, consistent const, private vs public variable)
- Big architecture, folder structure, inheritance, design pattern
- Don’t be too emotional
- People only pay what they think you’re worth
- But you control their thinking
- We need to clearly define and communicating your value
- Key question to ask when evaluating your value
- What are my clients’ needs and how do I meet them?
- What is my unique skill set that makes me better qualified to serve my clients?
- What do I do that no one else does?
- What problems do I solve for clients?
- What value do I add?
- 「休息」和「打發時間」是不同的
- 休息時有一件「需要專注投入的事情」
- 休息時進行一個「相對沒壓力的挑戰與練習」
- 「休息」是另一種有目標的行動
外在獎勵可以令人集中，可以有效令人完成簡單嘅工作。
創意工作冇清晰嘅界定，冇單一嘅solution，外在獎勵令人視野更窄。唔可以改善創意工作嘅performance，有時候仲令情況惡化。
新世代嘅人力管理重點
Because we filp it horizontally when we present the item to mirror.
How to combat imposter syndrome
- Receiving positive feedback, cannot ease feelings of fraudulence
- Hearing that an advisor or mentor has experienced feelings of imposterism can help relieve those feelings
- We can combat your own imposter syndrome by collecting and revisiting positive feedback
- Think about how much go to ourselves when success. Avoid ignoring the luck factor and over-estimate our skills.
- Quantifying our thinking.
- Train ourself think in number.
- Try to use more precise words, or speak in numbers
- Gut feeling good at handling everyday-things we have lots of experiences
- Slow and careful analysis is better for big things
- 3 Interesting memes by the speaker
- Success is sweetest when you achieve it across a large sample size.
- Your gut is your friend and so is a cost-benefit analysis
- The future is unknown, but you can damn well try and estimate it.