# 20210???_retrospective
## Improvement
- 教學錄影壓縮以後，檔案超小（都是靜態投影片），應該需要在某地方寫怎麼留存 tech share

# 20210618_retrospective
## Good
- 越睡越多，還睡過頭（ＨＨ）
- 研究 GPU 加速有進展 覺得開心 (Howard)
- 最近問 Bill 很多問題，讚讚 (Roy)
- content QA 已經有在定義了，越來越多場景看起是正常的 (hao.kai)
- 宮廟 Spatial mode 終於做完雖然不是我做（hkazami）
- 居然有人要幫忙商業模式問題 (hans)
## Improvement
- 幫攝影師組電腦可以做得更好 (Howard)->下次給測試影片然後讓攝影師填表格，會減少往返次數
- techshare, 算法 review 應該要錄影，這樣就不用重新介紹給實習生跟新人 (hao.kai)
- 每天都不知道昨天在幹嘛，應該再來寫 work log（hkazami）
- Tower 做功能的時候 spec 感覺開得太模糊，既難追蹤也容易浪費開發時間，應該早點釐清（hkazami）
## Impediment
- 影像處理完呼叫 tower 做 integration test 會讓 repo 一直長分岔 (hao.kai) -->不是問題
- 體感互動需求令人害怕 (hao.kai)
- 最近雜事很多無法專心寫 code (Bill)
- chefu 說想離職
## Discussion
- 阻擋購物中心的需求 (chefu)
- preplanning 加上 BD 更新
- chefu 會看構想的 slides
- 體感互動需求令人害怕
- 先不處理還沒成為議題
- 展示衛武營音樂廳在衛武營外面，用體感觸發
- 台北車站用體感旋轉體球
- implemenation 常常跟主管需求不符合直到實作完成時
- 寫什麼程式
- 來回往返，調整透明度、介面 -> 寫計畫書（不可能）？ mockup ? spec ?
- 設計師太少？
- tower 是內部工具，不會問設計師，但理論上是可以提需求的
- 設計師 實際上是 tower 的使用者
- [use this for prototyping](https://moqups.com/)
- 就算沒有人畫設計，自己先畫留下文件拿來討論 (stefan, liz, danny 看)
- 主管太忙會比較嚴重，大家需要等待
- 拍多一點影片給大家影片，來回往返速度會變快 VVVV
- 怎麼寫程式
- feature 可能太大，這是舊問題
- 不知道寫 code 要順著架構寫，還是改寫架構
- 嘗試畫架構，寫 function，討論後，調整架構，再開始寫
- 管他的鮮血一版 b>z，不影響 performance 的先寫下去
- interaction 東西比較少範例參考 react hook 寫不好容易重複 render
- reform OKR
- [research](https://hackmd.io/@falontw5566/SkuE0XfOd)
## Action Item
- 07/02 商業模式、roadmap proposal hans
- 做 tower 的 FAQ bill
- tower trouble shooting 值日生機制 hans
- 資料錯誤的影片回傳給負責人 hsiao hsuan
- 下次紀錄sprint review 超時時間 bruce
- 實作 refactor 前置溝通 (Hans)
- 生測試手機（bruce, hao.kai）
- 有列清單，還沒提申請買，要提出申請 Bruce
- 六月討論 reform OKR 是否可行 (Hans->Stefan, Che-Fu, Hao.Kai, Bruce->Ariel)

# 20210604_retrospective
## Good
- Tour mode 令人感動 hao.kai
- Chefu 有幫忙釐清項目讓開發比較順利 hao.kai
- 解說 python coding style 好像有幫助 hao.kai
- 感謝哲輔協助尋找 texture 包在單檔案的模型格式 - Bill
- 感謝 bruce 幫我撈影片和當與工讀生間的橋樑 YH
- 要加速後開始對每個指令斤斤計較 YH
- WFH 可以變瘦 (Roy)
- 心凱 3D 模搞得莫名快，覺得療癒 (hkazami)
## Improvement
- cuda 教學應該要早點辦 hao.kai who are !!you YH
- 邊討論邊做 Design doc 開發速度好像比較快 hao.kai
- tour mode 讓人崩潰 (Roy)
- 工讀生發燒，抖（HH）
- 對 tower & tter 的功能和 code 其實不熟（hkazami）
## Impediment
- Hans 設計的測試架構沒看到呀，也沒有實行 hao.kai
- 不知道調色接下來要跟誰合作 (Bruce, chefu) hao.kai
- 建spatial mode 還是滿困難的，source marker 跟 tower repo 上的label 也都還有誤植，都得再三回 tower 上看影片確認，換到nas12上跑行為也不太一樣ㄌ（HH
- 真的沒時間寫 code，約 80% 時間都在開會 (hkazami)
- Sprint Review 時間要掌握 (bruce)
## Discussion
# 檔案傳輸斷掉檔案
- 如果傳輸很小檔案，pod 重開
- QA 流程
工讀生人力不夠，立即的 bug 放到 jira board
functionality sheet, content sheet
- 奉天宮、鎮瀾宮 source marker 有些錯誤
弄好表格回傳給負責人 Bruce，根據每次的錯誤建立規格

- 大家現在真的有比較知道自己做的 part 是幹嘛用的嗎 (hkazami)

# 大家有更清楚做什麼事嗎？chefu
Hans: 不知道兩個月後會做什麼，東西長什麼樣子
hao.kai: 差不多，但是時程更清楚

## Action Item
- 資料錯誤的影片回傳給負責人 hsiao hsuan
- 下次紀錄超時時間 bruce
- functionality sheet, content sheet 貼到 smart-city hans
- 溝通 refactor 範圍 (Hans)
- 阻擋購物中心的需求 (chefu)
- 生測試手機（bruce, hao.kai）
- 有列清單，還沒提申請買，要提出申請 Bruce
- 六月討論 reform OKR 是否可行 (Hans->Stefan, Che-Fu, Hao.Kai, Bruce->Ariel)

# 20210521_retrospective
## Good
- 謝謝大家 下台一鞠躬(YiHsuan)
- 小島工作94爽 (Bill)
- Tower python api 很好用(hao.kai)
- 3D object is back (hans)
## Improvement
- debug from home太累，應該要review code(HH)
- 要在 scrum channel 上下班呀(hao.kai)
- 不要ＸＤ為了儀式感的話就可以（hkazami）
- 不知道大家在做什麼常常在狀況外(hao.kai)->O不夠明確
- 白白寫了一個 Tower python api (hao.kai)
- 大幅refactor造成幾個feature卡著，造成branch好大支(Roy)---> A,B,C feature 合併回 D，要解3次 conflict
## Impediment
- 突然要做 product management 了，老闆罵人就不能事不關己很不方便（hkazami）
- Catch-up 需要時間，有時候很難了解事情為何發展至此，或為何會卡死
- 更少時間做正事
- 家裡網速太慢，傳輸影片很久（YH)
- 儲存空間只剩下 15TB, ceph 不能存了（hao.kai）+1 (HH)
- SfM repo 乾淨程度有疑慮 (hao.kai) -> hh 會看 code
- 傳檔案 vpn 斷線，看不到實驗結果 (hao.kai)+1(Roy)
- 安全取得食物的管道(hao.kai)-->學煮飯、問yh
- 突然發現椅子很重要(Roy)
- hkazami +1 而且椅子大缺貨已經買不到了
- 桌子也很重要 (腕歲道爆炸, Howard) --> 左右開弓
- Beatus 心情貌似很差，易軒 wfh 她炸裂（hkazami）
- 媽祖繞境影片 release 沒有 QA 也沒有按照 tower pipeline commit tag（hao.kai)
- 被refactor卡住了Q (Bill)
- 購物中心聽到就怕（hkazami）
## Discussion
- Refactor 怎麼做比較好(為什麼突然做大 Refactor？)
- query 速度太慢 這是真實需求
- refactor module 要切
- refactor 範圍切 phase
- 多一個人可以 merge？hkchen 認為沒用
- 討論接口跟誰先上無關，都需要討論
- stefan 有通知 infrastucture 修改
- module 要寫有彈性一點
- 請想要改動的人要預估時間？
- 儲存空間只剩下 15TB, ceph 不能存了
- morale event -> 沒有勇者吵架
- 長期 PM 、過渡時期模式
- 短期
- 建立好 framework chefu 就離開
- chefu 跟 bruce 溝通這只是暫時的
- 長期
## Action Item
- 溝通 refactor 範圍 (Hans)
- 阻擋購物中心的需求 (chefu)
- vpn斷線統計表格（HH）
- 推動、觀察 QA (hans)
- 生測試手機（bruce, hao.kai）
- 有列清單，還沒提申請買，要提出申請 Bruce
- 六月討論reform OKR 是否可行(Hans->Stefan, Che-Fu, Hao.Kai, Bruce->Ariel)
- 有請Bruce也買一本OKR聖經本來看，Hans已經看完。預計近期與Bruce還有哲輔浩凱討論形式與實際作法，然後結案後分頭進行說服。
- 飲水機（Roy）
- 過了三天就WFH了，pending

# 20210507_retrospective
## Good
- docke image multistage build technique. hao.kai
- brcue 有釐清 face detection 需求 hao.kai
- 久違的寫 code 雖然只是 script（che-fu）
- 可以搶到很多 GPU 很讚 (Howard)
- 媽祖扭頭很療癒(HH)
- 從工具之王Jordan那學到很多好用的tool (Bill)
## Improvement
- face2face 跟 engine team 討論比help desk有效 hao.kai
- bruce 說大家最近比較常遲到 (hao.kai)
- 沒有 tower auto pipeline (hao.kai)
- 有點久沒sync tower以及cli的東西了 (Bill)
## Impediment
- 知道了怎麼用 vscode 進 pod 工作，覺得方便但 engine team 說會搞爆 ceph 覺得哀傷（che-fu）
- "ranger" & "k9s"
- gpu 相關的 build code 建立環境比較花時間很絕望 (hao.kai)
- gitlab ci OOM by "make -j" hao.kai
- face detection 插 ticket 花很多時間 hao.kai
- air condition is pretty bad in the office(HH)
- 公用電腦 Storage 沒有裝好 浪費到 Bruce 時間 (Howard)
- 想到要加速 gopro max stitching 200 倍覺得很絕望 (YH)
- 在家review code發現gitlab爆炸了 (Bill)
## Discussion
- NFS
- Tools
- use k9s to switch between pods
- ranger
- 跟大家正式說明 hsien 轉 team 的事情（che-fu）
- 下次要先跟老闆討論
- 考核結果自己不知道
- 前主管、人資、主管討論以後的管理？不好
- 告知人資要通知現任主管考核結果<-要做
- gpu 相關的 build code 建立環境比較花時間很絕望 (hao.kai)
- 問看看醫學影像的人都怎麼弄的
- 多踩雷
- Morale
- ~~~十二籃粥火鍋~~
- 日暮裏(火鍋)788 +2
- Belini Pasta(義式餐廳)500 +1
- 和牛涮(壽喜燒) 738 +7 :drooling_face:
- 菜豚屋 +4 (qhan:cry:)799
## Action Item
- 告知人資要通知現任主管考核結果 (hao.kai)
- 開一個 nfs 共同空間 (hao.kai)
- 觀察五點前空氣不清新狀況（hh）
- 更新測試文件(hans)
- 第一次測試流程起跑
- 穩定性的測試格式（系統的人）
- 有請大家寫了，之後慢慢補上
- 生手機（bruce）
- 六月討論reform OKR 是否可行(Hans->Stefan, Che-Fu, Hao.Kai, Bruce->Ariel)
- 有請Bruce也買一本OKR聖經本來看，Hans已經看完。預計近期與Bruce還有哲輔浩凱討論形式與實際作法，然後結案後分頭進行說服。
- 飲水機（Roy）
# 20210423_retrospective
## Good
- 學到 numpy, pytorch, cuda stream 快速寫法(Hao.Kai)
- 猜到 gopro max xxx.360 的投影(Hsuan)
- 導覽模式經求救與同仁們討論較有進展，開心：）(Roy)
- geojson editor變很好用～～（HH）
## Improvement
- 其實有目標但不知道目標會不會被管理(hao.kai)
- tower staging 更新後不會通知,新功能要自己通靈(Cindy)
- 不知道要怎麼跟工程組的人合作(Cindy)
- 合作開始的時候，確認名單，然後約好聯絡時間頻率，還有到 merge 前要彼此通知
- traveler 沒有統一的測速工具(Cindy)
- 實際上的問題是：不知道怎樣在 traveler 上測速
- retro 上討論的事如何確認其結論(Roy)

## Impediment
- 事情很多，牽涉範圍廣，上面第11點有點難以決定名單(Bill)
- 快要去看眼科了(YunHsuan)
- tower-staging 一次 commit 太多repo build 太久了，而且commit 也會失敗(HH)-> 用 production
- 沒有新PM，結案報告海(Bruce)
- Discussion單一主題太久(roy)-> 20min左右主提要另外約
- 不太知道工程組的進度（chefu）
- 被默默遺忘的 task force

## Discussion
-  玻璃門事件
- 請遵守時間，如果會議時間延後，要通知其他人
- hh 是會議主持人，去叫人
-  測試流程
traveler第一個禮拜測穩定的，第二個禮拜測新code
- OKR meeting is not OKR at all
- O can be killed
- 大O: 賣到國外
- 團隊重要目標被置換（自由->導覽）

## Action Item
- 更新測試文件(hans)
- 穩定性的測試格式（系統的人）
- 生手機（bruce）
- 快速寫法秘笈(Hao.Kai)
- 六月討論reform OKR 是否可行(Hans->Stefan, Che-Fu, Hao.Kai, Bruce->Ariel)
- tower： add tool tip(TOWER TEAM)
-
# 20210409_retrospective
## Good
- peaceful(HH)
- 終於有測試機了！(Bill) (Roy)
- TT QA 架構有逐漸成形了(Bill)
- 放假開心(yi-hsuan)
## Improvement
- ticket在planning時尚未釐清，估量loading效果不彰(Roy)
## Impediment
- 我不知道導覽模式是怎麼出現的（hao.kai）-> 現狀是導覽模式最重要，danny 設計的
- 做某一項測試做不出來，不確定要不要做下去（howard）
- yun 沒有準時做投影片，我常常要 review 幾天 (hao.kai)-> 下次早點
- 面試的時候 demo tter 失效(hao.kai)->不要 demo，以後有 QA 就好
- 不清楚AILabs blog的requirement. (Danny)
- gitlab traveler repo permission(yi-hsuan)
## Discussion
- 做某一項測試做不出來，不確定要不要做下去（howard）
需要哪個 function 做不出來 ->
討論這個 function 到底做不出的出來 ->
if hao.kai 跟 howard +專家的共識是不可能->停止 else 繼續做

可以列出嘗試過的方法（howard）
多多詢問為什麼做不出來（hao.kai）

- 不清楚AILabs blog的requirement. (Danny)
- requirement就是要讓其他公司的算法工程師或是學生看完文章後可以學到此專案要點，確保這些要點有被強調出來。
- 先跟mentor討論文章架構
- 訂完所有大標小標之後再把內容塞進去會最快
## Action Item
- 做某一項測試做不出來，不確定要不要做下去 (Howard, hao.kai)
- 探勘怎麼建立測試流程多種手機、瀏覽器版本報告給大家（Hans）
- 買手機（Hans 找 Bruce）
- content QA 測試格式（Hans, Arial）
- code QA 測試格式 (Hans, Stefan, Bruce)
- https://ailabstw.slack.com/files/T50G0CL3U/F01UB798EBA?origin_team=T50G0CL3U
- 怎麼建立測試計畫(Bruce)
- 統計手工跟電腦自動的每個影片處理流程所花的時間（Bruce）
- hao.yu 找 bruce 看 core ticket
- internal sprint review 分組名單(Roy/Haoyu)
- Biweekly meeting, 禮拜四下午 30min * 2 場 meeting
- 禮拜四早上 Engineering team 準備 outline
- Engineers 分成 3D, quality 兩組和算法 sync
- 若 Engineers 沒有更新，早上通知大家會議取消
- 加入 life in smart-city 教條: repo 權限找相關人士(HH)
-
## Good
- 真的開始為demo生content才發現tool的設計有需要改善的地方，進而可以討論跟改善，我覺得是好的(HH)
- TOWER
- GPS editor

## Improvement
- 影片處理流程不清楚，也不知道誰負責想流程（hao.kai）
- chefu 有想一些 cindy, jojo, chefu, ky 都有討論
- 用手工幾乎沒辦法做好
- 第一次可能沒辦法避免這個問題
- QA脈絡不清楚，解完自己手機上的bug也不確定其他device會不會有問題 (Roy)
- 因為流程上不順而卡來卡去造成效率低落工時長，壓力也大，還要瘋狂加班(HH)
- TT開始變龐大了有點難以開發，常常改了A然後B就爆了 (Bill)
- 原始影片是跟調色、source 長成兩個 branch （KY）-> hao.kai, chefu 沒有跟 PM jojo, bruce 確認上傳流程
- camera removal 沒有寫切頭切尾的功能，突然要加入（KY）-> 之後多一個切影片的流程
- source marker 錯太多（KY）->jojo會加兩個人檢查
- 沒有標準QA流程，感覺上只有看到什麼問題解什麼(Bruce)
- tter.cc 上的影片應該要完全ＯＫ才會上？（YH）
- - video quality QA jojo
- 借東西要告知一下（YH）
- Core tickets 應該要被追蹤 (Howard)
- 是不是太晚開始處理影片？ (Howard)
- Spatial mode 路線規劃應該再重新設計(HH)
- 目前頗亂而且邏輯不明(i.e. 3 or 4 秒的video or 諸多重複的路線)->jojo 有加一個流程，場勘以後會修改規劃，讓短影片變少，重複路線以後也沒有
- 服務監控的時程與規劃 (Jordan)
- - 問 stefan，這個時間點做這件事情有沒有意義？
- 空拍影片沒有gps要補上去是顆隕石，這個時候 must have 需求被解完（Hans）
- 列出 priority，做更優先的
- 問可不可以晚n天做出來
- 影片的來源(拍攝)能否早於demo前至少兩個禮拜以上 (Bill)
## Impediment
- OKR 會議接近 demo 會被取消（hao.kai）
- 一直無法取得蘇導演 8k 相機（hao.kai）
- Internal sprint review 忘記分組（hao.kai）
- 缺乏測試機 (Roy)
- 無法想像 tower 原始設計理念（hao.kai）
- 全部做處理
- git-- 問 jacobb
- hsien 遇到工作流程上問題但因為沒有來 retrospective 所以需要獨立解決（hao.kai）
- 該不該測蘇導演的影片 (Howard) (硬體)-->拿到相機在測試，因為可能永遠拿不到
- 有隕石(Hans)
- 在tter上看到影片有問題, 不知道這支影片是tower上的哪一隻, 花很多時間trace (cindy)
- url 前面是 stage_id 後面是 repo_id，按左下角也會顯示
- sudo merge(hh) -> hao.kai chefu 討論中
## Discussion
- morale event
- 唱歌 3
- 溜冰 1
- ikea 逛街+吃 3
- 樂盒子[4pm onward] 6
- 4/1(四) or 4/7 (三)

### 怎麼測試

Hans

QA測這個 sprint 出來的東西
daily build 每天測試一樣的項目，
integration test
PM先測試，之後丟給QA做全面測試

### tter看影片
按playground左下角的url複製鈕可以拿到該影片位於tter的url
假設有部影片url長這樣
https://traveler-playground.corp.ailabs.tw/region/sf/6046da5edd7a8f001335061a/6051a73d07c21d00ffc1bd36
前面那串6046da5edd7a8f001335061a就是stage id
可以貼到 https://tower.corp.ailabs.tw/stage/xxxxxxxxx 來檢視該stage
後面那串6051a73d07c21d00ffc1bd36就是footage id
目前footage id是完全照抄repo id的
可以貼到 https://tower.corp.ailabs.tw/repo/xxxxxxxxx 來檢視該stage

## Action Item
- 探勘怎麼建立測試流程多種手機、瀏覽器版本報告給大家（Hans）
- https://ailabstw.slack.com/files/T50G0CL3U/F01UB798EBA?origin_team=T50G0CL3U
- 測試報告這個屬於實際操作細節，實際開始Q之前寫好
- Video Quality QA (hao.kai, jojo)
- 怎麼建立測試計畫(Bruce)
- check video quality on tter.cc (ask jojo) (HaoKai)
- 借東西要告知一下(everyone)
- 統計手工跟電腦自動的每個影片處理流程所花的時間（Bruce）
- hao.yu 找 bruce 看 core ticket
- resume OKR for march(Bruce)
- internal sprint review 分組名單(Roy/Haoyu)
-

# 20210312_retrospective
## Good
- HDR bye bye ＾＿＾
- Side project-360影片比較 (Bill)
- 受到Hans大大跟Bill大大的幫助(HH)
- Sugar High (Howard)
- 內參校正有時間驗證 (Howard)
## Improvement
- 會議前先想好key point，直接切入重點 (Danny)
- HDR bye bye Q_Q
## Impediment
- 雨衣不知道可以放哪裡 (Roy)
- 不清楚什麼時候可以找到工讀生 (Roy)
- 新 PM 還沒進來 (Bruce)
- Release 時程 and Scope 不確定 (Bruce)
- 不知道老闆胃口 (Bruce)
- 不知道氣泡水要由誰負責補充 (Jordan)
- 不知道怎麼看 rectileaner projection (同tter.cc) 的 8k 影片 -->用桌機(hao.kai)
- RTK GPS/Video data is not very informative, has to pick it up
## Discussion
- HDR 何去何從 (Haokai)
## Action Item
- 下次下雨問雨衣要放哪#random
- Haokai 傳授 OKR 流程給 Bruce (Haokai, Bruce)
- Haokai 跟 Cindy 討論 HDR (Haokai, Cindy, CYY)
# 20210226_retrospective
## Good
- 春酒復活! (Bill)
- planning 時訂出 tickets 很好(HH)
- 有跟 Stefan 討論 requirements 很棒 (Howard)
-
## Improvement
- 10樓的咖啡貌似有點苦? (Bill)
- 有些ticket 沒法做因為沒有討論(HH)
- GPU 不夠, 訓完 model 或測 videos 要等一兩天才看得到結果, 有時候會不知道要幹嘛 (Cindy)
- 會議太多，時間零碎，效率很差 (hkazami)
- 零散的 Tickets 一直忽視關不掉 (hkazami)
- Plan 的題目太多，上一個題目還沒結束就開下一個 (Howard)
- 在產品上有蒐集密集路線，但卻沒有被放到 Playground 測試 (Howard)
-
## Impediment
- 踩到蟑螂屋(QQ)(HH)
- 物資短缺（衛生紙，堅果）(HH)
- 豆漿 (hkazami)
- 電腦壞掉了QQ (Danny)
- Cannot create volume and snapshot my job(YiHsuan)
- Not having my own seatQQ(YiHsuan)
- old rosbag 不見惹(HH)
- 電腦好像還是有問題，雖然修過一次ㄌ(HH)
- Stabilization 沒有辦法更新，但卻被需要 （Howard）
- cepth儲存容量不夠，剩下的資料都需要花很多時間收集(hao.kai)
- 沒有面試人力(hao.kai)
- AID人太少？(hao.kai)
## Discussion
- 會議時間or內容沒有很明確，有些有寄信有些沒寄信
- 老闆會問算法的東西怎麼都沒有上產品
## Action Item
- AILabs google calendar invites by group(HaoKai)
- 如何加更多 GPU (hao.kai, Cindy)
- look for old rosbags(Che-Fu)
- 發現做不完就提早告知並且re-plan (Howard, HaoKai)
- 密集路線排進去ticket？(Howard, AID)
- 搬家到minio(HaoKai, Stefan)
- AID人力討論(HaoKai, Che-Fu)
- 算法的東西每個月跟老闆sync一次(HaoKai)
- read through core ticket before next retro meeting(everyone)
- https://docs.google.com/spreadsheets/d/119PnxCZ8i_5fc0W7W4-OtiCeGgPPCZL8WcslrdxN_fo/edit
- Morale event
- 燒肉+火鍋 -> 原本家(雙連站) -> 6
- 做甜點(蛋糕) -> 7 -> win !!! -> Howard
- 頤宮 -> 3
- 樂盒子 -> 4
# 20210205_retrospective
## Good
- 可以修爛掉geojson 的editor終於初步released (Hans)
- Have time to edit product document and follow the milestone to solve problem (Jeffrey)
- Thanks for everyone's contribution (Jeffrey)
- Happy Chinese New Year and start a vacation (Jeffrey, Cindy)
- Implemented an algorithm (Danny)
- Finally done with spatial mode videos (hkazami)
- Finally discuss about planB of camera pose (HH)
- 去逛了新辦公室獲得第一手資訊(Bill)
- 好好檢視 TT core requirements (Howard)
## Improvement
- 希望一些重大里程碑的最小需求可以明確被寫出來-北車(Hans)
- run code without check the video source at first (YH)
- source was labeled wrong
- engineer haven't checked but believe the label
- Implemented an useless algorithm (Danny)
- do not understand ticket scope clearly at first (Roy)
- Too many meetings & slides (hkazami)
- Tower 經過這一陣子的洗禮？當初的架構設計已經不符合現今會使用到的範疇，目前需要重新設計了 (Bill)
- How to evaluate algorithm (Howard)
-
## Impediment
- It seemed the retro can't be finished in 2hrs, should we start the meeting earlier? (Cindy)
- 之後是否能訂個 feature deadline？ (Bill)
- 蟑螂家族 is growing (HH)
## Discussion
- Retro 要不要早點開
- 所有上次 discussion 後來放著沒討論的東西
## Action Item
- write down DoD of JIRA ticket(Roy)
- 需要有重大里程碑的最小需求(Jeffrey/Bruce/Jojo)
- Write down work log everyday (hkazami)
- 需要定義產品各個 release 的內容＆時程
- 要取得 PM, BD, 工程師, designer 等人的共識
- 內容與時程必須明確，且不能隨著外在交付時程的變化而任意更動
- 如果retro產生重大討論不完的議題，開ticket assign 相關人士
- maybe check our status after 1HR@Retro meeting
# 20210122_retrospective
## Good
- so far so good(Bruce)
- learned how to distribute work(Jeffrey)
- Bruce will take over the algorithm part
- spend more time on tracking schedule
- very productive(HaoYu)
- product requirement refined
- code reviewed by Stefan(Qhan)
- two days off(Chefu)
- AID collaboration meeting(Chefu)
- +1(Hsiao-Hsuan)
- no pitfalls(Bill)
- more patient on ethic-related topics(HaoKai)
- boss has constructive opinions yesterday(HaoKai)
- being very productive because of new screen!(Hans)
## Improvement
- hard to estimate run-time on different machines(YunHsuan)
- need more GPUs
- PM did not sync dependencies for different roles
- loading too high to do that
- e.g. 3D reconstruction/3D model scope
- no one has whole picture of product
- Product requirement(milestone) should be fulfilled really
- GPS editor developer does not seem to know what is the purpose of this tool(HaoKai)
- should fix GPS unstable points (Hans knows)
- footprint and little map
- Liz and Danny?
- should confirmed
- need TT internal review(HaoKai)
- regular meeting should not be skipped(HaoKai)
- sprint review
- OKR meeting
- spent too much time on preparing sprint-review slides(Roy)
- 3 hrs(HaoKai)
- collect data daily
- GPS editor sucks! but has been solved(Hans)
## Impediment
- there is some wooden debris on my desk(Cindy)
- front-end design not yet ready(Bill)
- not having mindset for whole picture of product(Bill)
- self-evaluation form???(Bill)
- +1(Hsiao-Hsuan)
- trying new form of evaluation system because of the growth of company

- sudden new feature(3D icon) from Stefan during daily scrum(Hans)
- 3D icon has been forgotten
## Discussion
- GPUs
- Why algorithm is not ready? from boss
- How to solve sync problem
- sync milestone with boss
- mindset of whole picture
- sudden new feature(3D icon) from Stefan during daily scrum(Hans)
- 3D icon has been forgotten

## Action Item
- Ask Beatus about 160NT event(HaoKai)
- learn vault and teach everyone(YunHsuan)
- come up with plan B for camera poses(AID)
- communicate with Stefan about his decision making process about only 4 gopro videos on demo(CheFu/Jeffrey)
- Prevent content resources are totally replaced by someone.(HaoKai)
- check if Liz will continue UX/UI design(HaoYu/Jeffrey)
- ask jacobkk about estimate run-time(YunHsuan)
- keep regular meeting, chat with Stefan(HaoKai)
- If Arial will not attend sprint review, Jeffrey will inform HaoKai(Jeffrey/HaoKai)
- ask Beatus about wooden debris(Cindy)
- Jeffrey should fix UX design problem(Jeffrey)
- Create an anonymous channel for employees and Ethan(CheFu)
- lunch 羊肉爐 morale event

# 20210108_retrospective
## Good
- code merged!
- setting time slots for HDR video
- tower is more like a product
- Jeffrey talks more details during scrum
## Improvement
- being late because of leaving too close to office
- speed-up rotation in 360 Utils
- should make sure function has been used correctly
- should prepare a clean env to test function efficiency
- meeting 1/8 deadline is pretty problematic
- no plan B
- no checklist
- scrum of engineering team takes too long
- how to write test cases for CV algorithms?
- create list of test images(videos) to be checked
- pytest(only passed or failed)
## Impediment
- can not assign for year-end party
- door of meeting room is broken
- how to ask people to work overtime
## Discussion
- issue report JIRA board in engineer slack channel for lab-engine issue
- meeting 1/8 deadline is pretty problematic
- no plan B
- no checklist
- product milestone
- plan B
- should write stories for product milestone
- JIRA board for product milestone
- not being able to plan
- explosive scrum event
- scrum length of engineering team
- morale event
- 羊肉爐
- 莫宰羊
- 哲輔家
- 溜冰
## Action Item
- Ask Beatus about 160NT event(CheFu)
- learn vault and teach everyone(YunHsuan)
- fix communication problem(HaoKai)
- 2021/01/14
- how to ask people to work overtime(HaoKai)
- come up with plan B for camera poses(AID)
- discuss product milestone(HaoKai/HaoYu/CheFu/Jeffrey)
- communicate with Stefan(CheFu/Jeffrey)
- Prevent content resources are totally replaced by someone.(HaoKai)
# 20201225_retrospective :christmas_tree:
## Good :smile_cat:
- GPU get! :smiley_cat:
- code reviewing process is faster
## Improvement
- spent too few time on video consistency
- spent too much time on tuning parameters
- ask people if struggling over 1 hr
- tower is under very high pressure
- see the problem but no time to fix it

## Impediment
- the U-shape seat is not comfortable
- desk is also too small

- feels pressure at the first week, but now it is better
- deadline is not defined for different stages
- pitfalls of videoIO

- lab3-g8 is not fully used
- no test cases for pipeline repo
- USB-hub makes me nervous
- ask for a keyboard
- can not benchmark on lab-engine
- very slow processing time (200-300s for one frame) - hard to fix the timeline of development
- hard to fix the timeline for development(PM)
- can only fix a range
- define dependency
- define scope

## Discussion
- intern for DIBR
- before committing code
- local files
- communication
- test case is not usefull
- too many redundant functions
- workload is not very clear to everyone
- everyone is not synchronized
## Action Item
- water machine project(WMP) (CheFu)
- Ask about 160NT event(CheFu)
- learn vault and teach everyone(YunHsuan)
- figure out where(or what) is the messy transformation (Hsien)
- before committing code
- run test
- check result
- collect the slow-processing cases on lab-engine(HaoKai)
- issue report JIRA board in engineer slack channel
- fix communication problem(HaoKai)
- 2021/01/14
# 20201211_retrospective
## Good
- become familiar with work flow
- code reviewing process went smoothly
- efficient week
- code get review, very fun
- 8 GPUs get!
- New screen
## Improvement
- not as efficient , due to sickness and other stuff
- many issues emerges during pipeline development
- issues has not been prioritized
- sickness
## Impediment
- Garlic is leaving QQ
- not so sure where to put doc in repo
- put layered README.md
- some additional work from certain request from PM/PR
- IT: Sean
- what is the scale of a experimental branch?
- only when some feature is fixed then merge into master
- Need a new screen but Beatus said there is no screen now
- getting sleepy because winter is coming
- Tower still need improvement
- DIBR has been forgotten
- Jojo has difficulties on guided tour mode but not many support from peers
- data collection for experiment is delayed
- need recruiting
- pro 1, camera car
- a bit surprised by the long list of videos
## Discussion
- Event driven support from Engineer?
- part-time worker
- IT: Sean
- ask HaoKai
- Jojo has difficulties on story mode
- with defined spec
- need Jeffrey's help
- drawing pictures and put them into Spec doc
- Dec. morale
- 無老鍋 [V] 12/18 12:00
- 辛殿（吃到飽）
- 泰滾

## Action Item
- intern for DIBR(HaoKai)
- water machine project(WMP) (CheFu)
- Ask about 160NT event(CheFu)
- learn vault and teach everyone(HaoKai)

# 20201130_retrospective
## Good
- separate scrum from TT and the rest
- improve efficiency by lock yourself into a small room
- DISC is fun
- OpenMP
- can ask Stefan
## Improvement
- should clarify motivation during paper-review presentation
- being late often, not efficient, being sick, too many side-projects
- too much freedom on jira tickes, still not very detailed tickets during planning
- somehow there is pressure
- can try to let people realize the situation
- SfM pipeline still have many pitfalls
- should mock first to test each modules
- Note for new-comers should be more complete

## Impediment
- need maintainer to merge gitlab code(under namespace SmartCity)
- need to ask for permission
- need to think about who's maintainer
- deploy token:
- vault
- Suddenly realize I have to go to DISC on Friday, lost one workday
- having double account on slack
- repo default become private
- Cindy not getting GPU
- just stay in line

## Discussion
- should we still need sprint rehearsal?
- optional
- if not attending rehearsal, should appears in outline meeting
## Action Item
- water machine project(WMP) (CheFu)
- Gantt chart critical path analysis(Jeffrey)
- see if screen can be exchanged(HH)
- Ask about 160NT event(CheFu)
- Sprint Review presentation note(everyone)
- Motivation
- result with comparison to show the progress
- with pictures
- DoD
- learn vault and teach everyone(HaoKai)
- add Life in Smart City to on-board note(Roy)
# 20201113_retrospective
## Good
- Tower website is released <3
- work for front end is ready to take place
- che-fu is early to work during this sprint
- more people being on time
- YunHsuan's code become much easier to read
- write a lot of code
- Hans write amazing backend
- follow Tower better
- can visualize user-case now
## Improvement
- Still lack some knowledge of the infra, there is some inconsistency
- add broader picture of the project in sprint review presentation
- can not catch up on hsien's scrum content
- with food?
- messy git branching style(do not make a metro map), maybe due to master not so mature?
- doing too much tuning...
- can consult with mentor to make sure if those extra work is needed
- actually...Cindy ask this parameter tuning...
- Gantt chart
- time line for engineering and data collection is not related
- critical path analysis
- 5 hrs on asking for robot car
- due to short sprint , progress is not as expected
- works more efficiently during late night
- not very efficient external communication
- talk with Ariel and Ethan way too late

- can not delete jira ticket
- just put it do done and close it
## Impediment
- The offer for cameraman is way too low
- data is not ready, there is problem with data collection, and no one is tracing the progress
- video for SfM is not with good quality
- video for SfM is way too long, much longer than expected
- Screen can not be rotated
- hard to describe the pipeline tuning progress
## Discussion
- offer for cameraman
- 160NT, this guy is treated as part-time worker
- actually what we need is walker or pilot
- how to make scrum more understandable
- can stop eating while talking
- sprint review format
- show more quantified data? motivation & DoD
- DoD(eployment):
- mazu group chatting
- internal user study

- git branching policy
- Never merge master to branch
- use git rebase before merge back to master
- can squash commits before rebase master
- open git branch should also consider dependencies

- sprint period is not aligned with monthly sprint review
- video length is 100X for SfM
- need a meeting
- robot car
- train buyer
## Action Item
- water machine project(WMP) (CheFu)
- Gantt chart critical path analysis(Jeffrey)
- Data collection for 3D SfM should follow good work flow(Jeffrey)
- Need to complete data collection items(Jojo/HaoKai)
- see if screen can be exchanged(HH)
- Ask about 160NT event(CheFu)
- Find solution for cameraman(HaoKai)
- Tell Ethan only update timeline during OKR meeting rather than sprint review(Jeffrey)
- Sprint Review presentation note(everyone)
- Motivation
- result with comparison to show the progress
- with pictures
- DoD
- Held a meeting for 3D scene list(Che-Fu/Jojo)
- describe flow of SfM image catching
- review 3D scene list
# 20201102_retrospective
## Good
- Cindy is very strong this sprint
- DISC is very useful
- should be more actively talking to Boss
- Somehow Danny learn about our project by himself
- Tower planning is good(yeah~)
- +2
- with planned tickets you can avoid unnecessary peer pressure
- sunmoon bike day is FUN~~~
- start to embrace javascript <3
- separate sprint planning can get to more details
- face production issues
- BBQ is sooooo delicious
- Get pumped after holidays~~
## Improvement
- The team are late more often
- no one told Danny about our project
- were late to office often recently
- job dependency problem
- structure of pipeline
## Impediment
- can not play video with Quicktime
- should convert file format
- mosquitoes in 10F
- waste some time on 3D model construction
## Discussion
- tower planning
- morale event
- 密室 +2
- 溜冰 +4
- 跳跳床 +0
- 爬山 +4
- 抱石 +6 +6
- [唱歌] +5 +6
- 泡溫泉 +3

## Action Item
- evaluate how many new GPU enhanced shared computers we need?
- algorithms(YunHsuan)
- designer(Jeffrey)
- write pitfall for videoIO(YunHsuan)
- add new rule for doc for new-comers(HH)
- water machine project(WMP) (CheFu)
- find out root cause of mosquitoes in 10F (Hans)

# 20201016_retrospective
## Good
- holiday~yeah
- schedule is useful
- 3D pipeline has first version!

## Improvement
- sprint review overtime(10+ mins)
- planned sprint goal is not being taken seriously(Tower)
- platform is still changing even at the end of sprint
- not well planned
- Reviewing process starts too late(stitching)
- pending for a week
- some discussions do not have all PICs, so the conclusion is easily get overwritten
- Stefan
- Ariel
- Danny
- Ethan
- Howard
## Impediment
- what is token is gitlab
- has to regenerate new token if forgotten previous one
- fix: just copy previous script
- how to do benchmark on babylon?
- performance issue
- can not preview video on Tower
- only 2 shared desktop
- 3D recon. flow is not scalable
- just make smaller models(for 3D env. models)
- Still PR job haunting
- easily get disturbed by delivery boy/girl
- sit too close to door
- many participants during scrum recently
- can not use personal laptop view 3D model

## Discussion
- rule no. 25
- remove it
- sprint review timing
- asking for more time
- policy: everyone should present
- tower planning
- tower story not defined
- no well defined goal for each sprint
- no user input for platform design
- no well defined use cases
- how to do benchmark on babylon?
- performance issue
- result validation
- 0,1,2,3,4 kings
- boss asking for bullshit
- as a result, it is ok this time
- FIX: if one of the king is going crazy, the villagers should sync with other kings.
- how to deal with personal packages?
- 6F?
- Too many people during daily scrum

## Action item
- open a pool for naming our platform(YiRu)
- evaluate if we need new shared computer(PM)
- asking more time for sprint review(CheFu)
- make list of use cases for tower(Tower Team)
- Separate BD/PR and engineers during daily scrum(Everyone)
# 2020930_retrospective
## Good
- so efficient! but i dont know why
- not because of tomato
- arrange 2 hr job while waiting for 1 hr job done
- script for submitting job ready
- many helpful people
- YY
- user cases has been defined
- work does not need as many debugging as anticipated
- learning git
- good way to do experiment
- working journal keeps good track
- tried every possible things

## Improvement
- solution is not applicable on our situation(big file)
- need to be more careful about new method
- no detailed tickets have been planned
- task will grow unexpectedly
- see if need to seperate planning furthur more
- planed too much work during this sprint(Yun-Hsuan)
- may be no work left?
- i want to train model

## Impediment
- 補班心累
- +1
- overwork for more than 10 days , illeagal
- no testing room
- siggraph paper takes 12 hrs than expected
- don't know when to call help from engine team
- test with two pods(one working the other not) and keep their hashes
- separated planning takes too long, and not even finished Cindy's part
- 1:30 pm ~ 4:00 pm

## Discussion
- no testing room
- make a convertable calibration chamber
- rotate object and white background
- 15F has a small room
- ask Beatus
- how to help jojo
- priority
- task management, no sense of the relations between tasks

- good way to do experiment
- use git commit ID to bind with experiment in/out-put and results
- what happened on CPL planning????
- should check with the newest version of work before sprint planning
## Action item
- estimate GPU/CPU request(HaoKai/Stefan)
- plan to make a convertible calibration chamber(HaoYu)
- Ask 15F room(HsiaoHsuan)
- try to do something with Jojo(Jeffrey)
- should check with the newest version of work before sprint planning(everyone)
# 2020918_retrospective
## Good
- meeting oftenly so errors can be fixed quickly
- design review is useful
- +1
- good to have discussion about use cases
- feel good for having test code
- finish MVS paper
- jacobkk is awesome
- know babylon better
- may need to ask people instead of reading all the code?
- working journal is effective
- 10F problems get addressed quickly
- Jojo and Jeffrey is doing their duties
- know what to do by translating docs
- but need to define what i want to do here
- need story for our platform
## Improvement
- waiting for job done, some kind of waste
- check every certain time
- did not allocate my job very well
- so tired
- no lunch for 4 days
- sprint review slides should be submitted on time
## Impediment
- too many meetings so no time for coding, but it is what it is
- my interview questions get posted on PTT
- people are hitting my chair
- get called-out during my honey moon
- QQ
## Discussion
- too many non-tech meetings?
- may not be a problem, let's wait and see
- morale event
- BBQ@Che-Fus
- 10/23
## Action item
- 3D reconstruction workshop(hsien)
- 9/21(after 3:00pm)
- estimate GPU/CPU request(HaoKai/Stefan)
- delete epic game engine (Hsiao-Hsuan)
- delete Tai61/Tainan/chingo(HaoKai)

# 2020904_retrospective
## Good
- Good to have CF to refactor my code(HH)
- refactor is good thing to do(CF)
- keep working journal is useful
- stitching project can be written by multiple developers
- define requirement statement
- get the project fund!
- +1
- no longer have to travel for scrums
- start to join scrum meeting next week
- talked with CYY, has been helpful
- more organized experiments
## Improvement
- refactoring is much more time-consuming than original planned
- working efficiency is slow down by coordinate transformation-related tasks
- fixed by writing test-cases
- not very clear goals still?
- for future goal, should assign priorities for tasks
- design review may not be useful?
- needed when getting to production
- project goals hasn't been very clear yet
## Impediment
- 10F office is worse than 6F
- drinking water problem(dirty)
- cold water switch is broken
- cockrouches around drinking area
- where to put wet umbrella
- air-con is too loud
- no toilet paper
- +4
- seat sanitizer
- landlord is kinda creepy :(
- bug in coffee machine
- slightly broken
- no enough hard-disk
- too many videos
- surveying how to store data
- someone is already talking about it
- no response after resource request(GPU, storage) at the beginning of the year
- GPU is really not enough
- GPU usage policy
- more GPU share machine like g2
- no backup for 媽祖繞境 video
- shared computers are now very messy, need some clean-ups
- no check for broken stored files?
- should be fine, data is safe now
- how to delete junk? du takes time
- ceph has function, need to doc into wiki
- defeated by babylon QQ
- CPU is hard to get as well
- only get very few cpus(~4 or 5)
- use ps to check if process is on
- Mac can cook egg sometimes
## Discussion
- GPU need
- morale event
- 橘色涮涮屋
- continue?(9/8 Tue) V
## Action items
- 3D reconstruction workshop(hsien)
- report 10F problems(CheFu)
- estimate GPU/CPU request(HaoKai/Stefan)
- backup for 媽祖繞境 video(Jeffrey)
- clean up shared computers and portable disk(Cindy)
-
# 20200821_retrospective
## Good
- product survey/use case study is done
- attending proposal meeting is a good learning experience
- time management is more efficient
- exploring new method
- Tower DB structure is OUT now~~~
- There is always someone there to stop pointless meeting
- Stefan is negotiable
- Good code with HaoYu
- Stefan helped with camera solution /HDR experiment
- max numbers in our history of attendance in morale event
## Improvement
- need design review before coding
- code review should only change logical errors
- annotator should have more efficient work flow
- too many parallel projects/testing running
- design on the wrong direction because there is no documentation for Tower DB structure
- did not defined what to design on the first meeting
- need something to remind myself of previous work
- keep working journal
- open questions
- to-do lists
- jira descriptions/comments
## Impediment
- pod got deleted by lab3
- dashboard is also broken
- is fixing now
- depth-map video for streaming? Tower?
- not to attack real problem, but to approach with some new methods
- Tower
- no very strong goal for what should be done
## Discussion
- design review
- class diagram after 1st working version of prototype
- how to plan a sprint?
- no pre-planning last sprint
## Action items
- 3D reconstruction workshop(hsien)
- design better work flow for annotator(Yun-Hsuan)
- Tower Documentation(Bill)
- clarify the problem of deleting pod(Garlic/hsien)
- class diagram after 1st working version of prototype(everyone)

# 20200807_retrospective
## Good
- god of babylon(巴神)
- houyu 神剪片
- find future direction
- communicate frequently
- knows who to find
- diffrent teams can cover up each other
- HaoKai is keeping good track of progressing
- review proposal
- time management
## Improvement
- garlic has too much load
- does not know how to help
- working flow not very efficient
- not very organized as result
## Impediment
- no clear deadline for requests
- communication problem
- lost track of priority of tasks
- schedule is ruined, totally
- new materials for demo
- rundown comes in too late
- one week before demo deadline
- too many versions
- HDR demo video version 1 was abandoned, and it takes a full sprint
- for whom? for which condition?
- scout fly return to the original version
- too many slack channels
- no specific time for discussion
- door-bell is too loud at 10F
- request keeps changing, and new request keeps comming in, have to change hard-coded stuff
- boss likes to throw a bomb
- Sunmoon
- MtAli
- FunTien with perspective camera
- mazu catch a bomb!

- no concrete demo format
- director version
- remains a mistery

## Discussion
- how many info should be disclosed from a meeting(OKR)?
- meeting minutes
- morale event
- group SPA(1500) +1
- pool BBQ(550) +6
- 8/21(Fri) evening
- jumping jack(270) +1
- to break your neck
## Action items
- 3D reconstuction workshop(hsien)
- make meeting minutes for OKR meeting(HaoKai, CheFu)
- try to fix specific time for deadlines and discussions (everyone)
- try to hide spared key of 10F office (Qhan)

# 20200724_retrospective
## Good
- Qhan has Garlic as mentor which is good
- Stefan communicate better with people
- Hao-Kai is an enthusiastic engineer
- PIC is controlling the situation during meeting
- peaceful
- Stefan finally give up some approach about 3D object trajectory
- good to have mentor
- air con. of 10F office is much better
- have time to fix some remaining issues
- Stefan has changed during this sprint
## Improvement
- not designing experiment well, lack of careful examination
- lack of experiences
- no design of user interaction mode on our platform
- UI/UX design
- content related design
- too many meetings
- combine retrospective and internal sprint review?
- not planning very well, and things get changed after planning
- tickets get stocked and have to be given up
- arrange meeting agenda to run the meeting more smoothly
- declare the purpose of meeting more clearly
## Impediment
- Yun-Husna has some kind of impediment?
- just lack of experience, not really a problem
- should confront openMVS parameters
- office is so far away(QQ)
- no sink in 10F office
- no water machine
- no trash bin
- temperature is not homogeneous in 6F office
## Discussion
- UI/UX design
- Ethan does not allow user study
- Stefan does not like guided thing
- we need UX designer
- combine retrospective and internal sprint review?
- keep it as it is for now, and check how it goes after two weeks?

## Action items
- explain to Liz why she should join sprint-planning(Che-Fu)
- discuss with people if spending to much time on non-tech issues (Hao-Kai)
- new time meeting:(everyone)
- 2:30pm internal spring review
- 4:00pm retrospect meeting
# 20200710_retrospective
## Good
- direction is correct
- B-day cake :)
- break-though in 3D model construction
- Wow, Howard is so awesome
- Wow, Garlic is so awesome
- wow, Garlic is so awesome
- good with more well-written/organized tickes

## Improvement
- Bill seems to be in trouble?
- forgot to mention attempt arriving time while being late
- different implematation for a ticket w/wo demo, confusing
- demo?? don't know what to do about 3D objects(quality/quantity)
- algorithm review
- too many people attented
## Impediment
- ceph is off frequently last week, but is fixed this week
- may be internet problem
- keeps kicking out from lab engin
- should report to engineering slack channel
- Beatus is not in good mood
- 權責不清
- working time(late? < 8hrs)
- double morale event
- Suddenly become someone's mentor?(Garlic)
- can not find doc sometimes
- change of right of making an appointment with office
- has been changed but without notification
- Scrum system is not working at all
- blur effect
- flyout effect
## Discussion
- tag in random while someone is late ????
- Hao-Kai is not gonna change
- whoever will be late should notify beforehand
- Beatus
- mood
- policy
- flow of algorithm review
- as usual is okay
- Scrum system has potential to fail
- morale event
- broccoli花椰菜（韓式餐酒館）400 +1
- 詠心奉茶 300-500 +3 (7/17 lunch)
- 無老鍋 600-700 +1
## Action items
- double morale event should ask permit in advance (ask for personal leave)
- add AID hackmd intro document (AID)
- light-weight intro for PM (Hsiao-Hsuan)
- add links
- doc for using "viewing video on-line" (Bill)
- assign reasonable work load for workers (HaoKai, Cindey)
- HaoKai assign, Cindy review
- meeting for PM/BD/Engineers/Stefan about demo(HaoKai)
- figure out why Beatus has bad impression on us(Che-Fu)
- preparing for design concern and propose during sprint planning(Bill)
- make notes during new feature request from Stefan(Bill)
- ask Liz to join sprint planning(Jeffery/HaoKai)
# 20200629_retrospective
## Good
- is more efficient for video taking with engineer reviewing quality of videos on spot
- demo director appears
- lucky
- many new things to develop
- my colleagues are so great! people wants to make things happen
## Improvement
- timeline can only be checked after 遶境
- scrum lasts for 30 mins
- features keeps changing (from demo director)
- sprint planning is not effective, feature will be changed during sprint
- intern spent more than expected time on 遶境
- get abused?
- have spent 6 hrs on discussion about new feature proposal
- how to introduce a new concept into a project?
- haven't make a single slides
- should create sub-task for stories
## Impediment
- clash between Ethan and Stefan, makes PM clueless
## Discussion
- how to plan without rundown/with demo
- how to introduce a new concept into a project?
- no agreed flow of work
- clash between Ethan and Stefan, makes PM clueless
## Feedback from last retro
- GPU will be priced until September, and the request will be checked from each team again
## Action items
- PM should use ticket system to maintain working items(PM)
- maintain intro document (AID)
- https://docs.google.com/document/d/1i6Jaj9NOGT9-iavc60C9QlYhcoVfN06aliRSmhFGtlQ/edit#heading=h.hgz81nmmjzot
- light-weight intro for PM (Hsiao-Hsuan)
- doc for using "viewing video on-line" (Bill)
- assign reasonable work load for workers (HaoKai, Cindey)
- HaoKai assign, Cindy review
- only key points should be mentioned during scrum(everyone)
- now the scrum meeting is too long
- if have personal leave, should put into calendar, and notify PIC
- add newcomer SOP to Wiki(Hsien)
- ex. add into smart-city google calendar
- inform PM the proper use of intern(Hao-Kai/Jojo)
- refine own stories as clear as possible after sprint planning(everyone)
- ask about demo rundown(Jeffrey)
- making meeting minutes and stop the discussion while necessary(Hao-Kai)
- arrange meeting for clashing (PM)
# 20200612_retrospective
## Good
- have someone to discuss with, many collaborators
- sprint review runs pretty smooth
- rehearsing is useful
## Improvement
- forgot demo deadline is on 6/12, and have to work overtime
- not so sure about demo materials, engineers are not informed
- does not hear from demo anymore?
- model runs too long, deep learning curve
- 遶境 takes too much energy
- worker get injured, overloaded QQ
## Impediment
- process was killed because of using too much memory
- recommended usage: 20G~30G
## Discussion
- new ideas? (one shot scenario)
- an official meeting should be held to discuss about a new feature
- PMs, designer(s), person who proposed, engineer representative(HaoKai picked)

- demo materials
- now we just submit our stuff
- but a rundown should be provided afterwards
- who will watch demo
- where will this demo be presented

## Action items
- PM should use ticket system to maintain working items(PM)
- sprint review slides(Jeffery)
- Outline on Monday afternoon
- Rehearse on Wednesday 14:00
- maintain intro document (AID)
- https://docs.google.com/document/d/1i6Jaj9NOGT9-iavc60C9QlYhcoVfN06aliRSmhFGtlQ/edit#heading=h.hgz81nmmjzot
- Ask for GPU resource(Titan) (HaoKai, Garlic)
- CK's feedback: virtual GPU
- where does number of Titan(4) come from? (Garlic)
- update GPU resource policy (Garlic)
- experimenting google-drive for cpl-docs (Hsiao-Hsuan)
- if not working, then switch to nas1
- prepare docs for common tools of the team (HaoYu)
- sftp
- viewing video on-line
- video IO
- utilities 360
- base images
- Find mentor for interns (HaoKai)
- assign reasonable work load for workers (HaoKai, Cindey)
- HaoKai assign, Cindy review

- when a new direction of project was proposed: (HaoKai)
- notice HaoKai first
- an official meeting should be held to discuss about a new feature
- PMs, designer(s), person who proposed, engineer representative(HaoKai picked)

- a rundown should be provided after by PM before 6/17 (WED) (Che-Fu)
- who will watch this demo?
- what kind of situation will this demo be presented?

# 20200602_retrospective
## Good
- timeline is ready
- knows who is overloaded
- smooth co-working experiences with other teams
- garlic is in <3 <3 <3
- sprint planning runs smoothly
- design of product is solid
- summary of product demo graphically (by Stefan) is good
## Improvement
- there's still some PM working items missing
- no ticket points
- some surprising new ideas?
- chu-hsuan is usually forgotten during stand-up meeting QQ
- Unfriendly environment for new-comers
- being late more often
## Impediment
- no shared GPU can be used
- need to assign GPU ID
- absence of stand-up meeting
- time of retrospective/sprint planning meeting is not align with sprint review
- unpleasant broadcasting from our building
- GPU job will be restarted without notification(due to wrong policy)
## Discussion
- GPU resource
- due time for sprint review slides
- Outline Mon. Afternoon
- Reharsal Wed. 14:00
- Should send meeting notice if any meeting should be held(everyone)
- assignments for intern(chu-hsuan)
- where to put tutorial materials
- experimenting ticket points?
- we can start putting real points on stories
- some surprising new ideas by Stefan?
- wait for final decisions
- stand up meeting policy?
- as usual
- Unfriendly environment for new-comers
- sftp
- viewing video on-line
- video IO
- utilities 360
- base images

## Action items
- PM should use ticket system to maintain working items(PM)
- sprint review slides(Jeffery)
- Outline on Monday afternoon
- Rehearse on Wednesday 14:00
- maintain intro document (AID)
- https://docs.google.com/document/d/1i6Jaj9NOGT9-iavc60C9QlYhcoVfN06aliRSmhFGtlQ/edit#heading=h.hgz81nmmjzot
- Ask for GPU resource(Titan) (HaoKai)
- Asking about GPU-job distribution policy (HaoKai)
- we can start putting real points on stories(AID)
- Join stand-up meeting virtually for remote-workers(all)
- Monday, Tuesday, Wednesday
- with daily scrum meeting Goole Meet link
- remind Chu-Hsuan(Jojo)
- Find a place to hold common docs for the team (Hsiao-Hsuan)
- Summarize work during each sprint with slides
- common tools
- tutorials materials
- prepare docs for common tools of the team (HaoYu)
- sftp
- viewing video on-line
- video IO
- utilities 360
- base images
- Find mentor for interns (HaoKai)
- New ideas should gain agreements between stakeholders(HaoKai,PM,Stefan)

# 20200508_retrospective
## Good
- have feeling of working together, smooth communication
- Jeffery doc many details regarding to work and they are useful
- meeting in general is more efficient(+1)
- get used to virtual meeting
- learn how to share a tab on google chrome
- better communications
- have opportunity to do product survey
## Improvement
- problem with sprint planning
- 3D tag? VO?
- need to prepare more about stories details before planning
- sprint planning becomes different than before
- JIRA ticket is not clear
- brain storming somehow not efficient and not conclusive
- too tech-oriented
- should be feature-oriented(ex: walk around)
- no narrowing down
- not conclusive at all
- do not jump to conclusion of some tech's application
- ex: 2.5D
- timeline is still not there
## Impediment
- user study
- PM and Boss not happy about doing user study
- use slack to discuss but not as efficient and open Meet need time(+1)
- WFH is less efficient than work in office when starting something new
- lack of planning with Stefan
- echo, lagged
- feels very unsupported(Bill)
- we don't have anything to follow, have to start a bran new project(Jeffery)
## Discussion
- feels very unsupported(Bill)
- dealine = June?
- man power
- feature?
- later than June
- may be better
- Chu-Hsuan can come in 3rd week of May (?)
- define Demo content
- deadline for sprint review slides
- Outline on Monday(3 days before) afternoon
- Rehearse on Tuesday(2 days before) afternoon
- Fix time-line for project(Jojo, Jeffery)
- ask for 5 mins time slot for each one during sprint review presentation (Jeffery)
- maintain a document for basic intro for tech and tech-feature correspondences(everyone)
- user study
- start with the features that most of people will appreciate
- test if some feature is effective for users
- do not jump to conclusion of some tech's application
- brain storming somehow not efficient and not conclusive
- set time bound of meeting
- set goal and meeting outline for meeting
- PIC
- goal of product is not clear
- timeline
## Action item
- Reminder: Meeting guideline (Jeffery)
- set time bound of meeting
- set goal and meeting outline for meeting
- Need a PIC (person in charge)
- sprint review slides
- Outline on Monday(3 days before) afternoon
- Rehearse on Tuesday(2 days before) afternoon
- Check Demo content and held discussion session to fix timeline(Jeffery)
- Should send meeting notice if any meeting should be held(everyone)
- make interns sprint presentation slides skeleton right after sprint planning and make them turn in final slides one week before(HaoYu)
- rephrase 3D tag part of sprint planning document(AID)
- put on earphone during virtual meeting(everyone)
- prepare a talk about babylon(Bill)
- maintain intro document (AID)
- https://docs.google.com/document/d/1i6Jaj9NOGT9-iavc60C9QlYhcoVfN06aliRSmhFGtlQ/edit#heading=h.hgz81nmmjzot
- talk with Ethan about user study together(Jojo, HaoKai, Jeffery)

# 20200417_retrospective
## Good
- WFH is smooth
## Improvement
- forget some meetings
- meetings last longer
- more than two people speaking at the same time during Google Meet
- The time for Sprint review is too short
- sprint planning is a bit fail
- stitching from side project
- topics that is not planned slip into the sprint
- no one is in charge of TOWER, should be Garlic, but no one knows what happened
- virtual meeting is less efficient for discussion
- sprint review slides from interns comes out usually late

- Jeffery used to start sprint review preparation one week beforehand
- the time-line for project is still not fixed

## Impediment
- FR should be maintained even during WFH period, should make some shifts for that, but Bill still need to be on-site.
- Stefan missed some discussion sessions of our project direction
## Discussion
- rule 15 (PM should put sprint goal after sprint planning)
- Jeffery will try to understand the sprint goal
- rule 23 (google slides)
- try go back to google slides and provide link of videos
- time regarding to sprint review
- TOWER status
- make interns hand in sprint review slides on time and how to check their progress?
- make slides skeleton right after sprint planning
- how to improve meeting process with virtual meeting?
- sprint planning is a bit fail
- maintain a document for basic intro of tech and tech-feature correspondences for PM to read
- Stefan missed some discussion sessions of our project direction
- people has different perspective and objectives
- should AID CPL plan together?
## Action item
- Set up JIRA account for Jeffery
- Should send meeting notice if any meeting should be held(everyone)
- prepare more before virtual meetings(drawing, documentations...etc) (everyone)
- sprint review slides should be out 2 days before review day(Jeffery)
- Fix time-line for project(Jojo, Jeffery)
- Negotiate with Zack about the remaining FR duties(Jeffery)
- try going back to google slides and provide link of videos, and presenter should take charge of the slides control for individual session, and remember to authorize your file. (everyone)
- ask for 5 mins time slot for each one during sprint review presentation (Jeffery)
- make interns sprint presentation slides skeleton right after sprint planning and make them turn in final slides one week before(HaoKai/HaoYu)
- maintain a document for basic intro for tech and tech-feature correspondences(everyone)
- make example for intro doc(above) (Che-Fu)
- audit CPL sprint planning(AID)
# 20200401_retrospective
## Good
- discussion with PM more often
## Improvement
- late in the morning
- no discussion about morale event
- low morale recently
## Impediment
- Information is not open between CPL and AID
- promoting WFH is slow
- late in the morning
- pre-sprint planning is falling apart
## Discussion
- who should review intern's code
- ask 心凱 to review chu-hsuan's code
- chu-hsuan review Qhan's code
- PM engagement
- PM feels no contribution during sprint planning
- PM feels frustrated
- Engineer should be more patient and explain more about tech details (constructive discussion)
- need a dedicate meeting and sufficient preparation beforehand for talking about ideas
- requirement from boss is usually time limited
- reject unreasonable requirement from PM politely.
- 360 camera data collection on car
- 8D9N
- pre-sprint planning is falling apart
- Stephan wants new tech (not within the scope of team project)
- Arial was bypassed in big project
- define topics of meeting and need a scrum master
- late in the morning
- attitude!
- precisely stating the delay of arriving time
- morale event in order to pump morale
- Has to make full-member attendance every two morale event
- 攀岩（室內）[3]
- 保齡球[0]
- 貓空[4]
- 漆彈/雷射槍[5]

- promoting WFH is slow
- make a post in general?
## Action item
- Mention ticket ID and title during daily
- Put making version number for common packages via git tag as a JIRA ticket (Hao-Kai)
- keep track with year/date...etc.
- ex. utility 360
- Add writing test cases into JIRA ticket and redesign if needed(AID)
- Sprint review presentation(everyone)
- clarify motivation and quantified result are important
- no need for too much technical details
- self-rehearse once
- have pre-sprint review meeting the day before slides due(CPL)
- fix close state(Che-Fu)
- build up the demo script (Jojo)
- explain how to build up the demo script to Jojo (Che-Fu)
- more code review for intern(everyone)
- precisely stating the delay of arriving time (if passed 9:30am)(everyone)
- intern code review(Hao-Kai)
- ask 心凱 to review chu-hsuan's code
- chu-hsuan review Qhan's code
- Engineer should be more patient and explain more about tech details with PM (constructive discussion) (Engineers)
- morale event: laser gun/paintball (Cindy)

# 20200320_retrospective
## Good
- code review is good (Cindy is good)
## Improvement
- code review (funciton side effect ) (QHan)
- too many extra tickets, short notice of demo
- the working time is chopped
## Impediment
- WFH
## Discussion
- explain side effect of a function
- use static function as much as possible
- do not update self member in functions, use a update function instead
- be careful of class variable
- how to make better time management according to type of work
- train intern to collect data
- how can we know when will there be a demo in advance?
- make Jojo notify us if there is someone important visiting
## Action item
- Mention ticket ID and title during daily
- Make version number for common packages via git tag (Qhan)
- keep track with year/date...etc.
- ex. utility 360
- Add writing test cases into JIRA ticket and redesign if needed(AID)
- Make office have open notification for WFH policy (Che-Fu)
- Sprint review presentation(everyone)
- clarify motivation and quantified result are important
- no need for too much technical details
- self-rehearse once
- have pre-sprint review meeting the day before slides due(CPL)
- fix close state(Che-Fu)
- fix the date with Jojo(Hao-Kai)
- train intern how to collect lidar data set(Che-Fu)
- build up the demo script (Jojo)
- more code review for intern(everyone)
- let engineers know if VIP is visiting (Jojo)
- ask if we can rehearse WORK FROM HOME(Che-Fu)

# 20200306_retrospective
## Good
- office provides us mask
- code review is helpful
- code review is pretty good
## Improvement
- work from home?
- code review takes time
- modifying slides takes time
- AID skip one sprint planning
- sprint review presentation is overtime
- too much time spent on reviewed code, since it is fundamental change.
- working schedule for annotator(s)
- sprint planning takes for whole day
- sprint review presentation is not straight forward for audience
## Impediment
- slides is overdue
## Discussion
- how to make ticket title?
- workflow for common utilities
- time well spent, there is learning curve
- what is reasonable time for sprint review presentation?
- clarify motivation and quantified result are important
- no need for too much technical details
- how to prepare for sprint review presentation(Cindy)?
- can not follow your own thoughts while preparing slides
- just follow what has happened if the sprint goal got changed
- how can make time restriction for sprint planning
- Arial joined designed review, should be fine now.
- working schedule for annotator
- create kanban tickets
## Action Item
- Mention ticket ID and title during daily scrum(everyone)
- Make version number for common packages via git tag (Qhan)
- keep track with year/date...etc.
- ex. utility 360
- Add writing test cases into JIRA ticket and redesign if needed(AID)
- Make office have open notification for WFH policy (Che-Fu)
- We should stick to the sprint planning and create JIRA ticket no matter what(AID)
- Change or split JIRA tickets title(sub-tasks) if needed and PM should be notified(CPL)
- Add "closed" status on JIRA board(Che-Fu)
- Sprint review presentation(everyone)
- clarify motivation and quantified result are important
- no need for too much technical details
- self-rehearse once
- have pre-sprint review meeting the day before slides due(CPL)
- create kanban tickets for annotator during sprint planning with excel(Cindy)
- PM engagement with engineers(Hao-Kai)

# 20200221_retrospective
## Good
- Slack daily scrum board is very effective
- less wantwant
- good
- professor is helpful
## Improvement
- AID need more careful test case
- More detailed about discussion conclusion
- only ticket ID was named during daily scrum
- package in dev. has no version number control
- not making much daily scrum meeting

## Impediment
- broken connector
- no professional front-end to help
- steep learning curve/mentor needed
- PM/BD discuss with engineer too often
- BD will ask intern directly
## Discussion
- Merge request should bind with JIRA ticket?
- need test case?
- how to write
- how to survive without professional front-end engineer?
- Stephan is too busy and can not review Qhan's code
- how to make the content of planned sprint tasks consistent?
- package in dev. has no version number control?
- e.x. utility 360
- BD/PM problem
- change of request: BD/BOSS/PM
- communication problem (PM can not convince BOSS)
- disagreement on dev. direction between PM and BOSS

## Discussion conclusion(can not be in action item)
- Arial should get more involved in meetings.

## Action Item
- Deal with the problem of decreasing reputation
- 補班, paid holiday(Che-Fu), 2 days a month work from home
- Discuss with ChuHsuan about the workflow and deadline(Hao-Kai)
- Mention ticket ID and title during daily scrum(everyone)
- Make version number for common packages via git tag (Hao-Kai)
- keep track with year/date...etc.
- ex. utility 360
- Ask for new broken connector with Beatus(Hao-Yu)
- Ask for front/back-end mentor for smart-city(Cindy)
- ask Stephan for mentor candidate
- front/back-end or a full-stack
- Jojo should use Hao-Kai if Boss need engineers opinion for dev. direction(Jojo)
- Arial should participate designer review meeting and pre-planned meeting(Hao-Kai)
- Merge request should bind with JIRA ticket(everyone)
- Add writing test cases into JIRA ticket and redesign if needed(AID)
- Ask who can be Qhan's reviewer(Hao-Kai)
-
# 20200207_retrospective
## Good
- peaceful
- making slides after each merge request is good
- work from home is good if you can only work independently
-
## Improvement
- HsiaoHusuan forgot to upload retrospective meeting minutes after meeting last time.
- run test before merge-request
## Impediment
- too cold.
- intern is late to work too often(missed daily scrum)
- intern changed ticket priority
- No one to discuss if research get stuck. Even consultant can not help.
- deadline is not reasonable due to state change of project
## Discussion
- reputation of AILabs?
- no clear direction
- benefit >>>>>>
- marketing/PR
- 360 video is not mainstream/exciting
- diverse projects
- flexibility & freedom of research

- how to remind action items

- Drone video processing merge problem
- CI/CD test?
- (too cold)stitch slipper?
- work from home
- what's wrong with deadline?
- intern is late/change ticket priority in mind
- no one to discuss
- morale event
-
## Action Item
- Remind PIC of action item via slack channel (Che-Fu)
- Think about how to get full stack engineer(Hao-Kai)
- Deal with the problem of decreasing reputation
- 補班, paid holiday(Che-Fu)
- Add action item as ticket to Jira(everyone)
- If a branch will last long, it should merge with master frequently.(look at graph) (everyone)
- Merge request should bind with JIRA ticket (everyone)
- Borrow stitch slipper from Cindy(QHan)
- Try 2 days a month work from home(Che-Fu)
- Discuss with ChuHsuan about the workflow and deadline(Hao-Kai)
- More detailed written scrum content (QHan)
- More review of algorithm (Hao-Kai)

# 20200117_retrospective
## Good
- no repeating rules is good
## Improvement
- sprint did not finish on time(AID), research related stories is bottleneck.
## Impediment
- printer is broken? PDF is not fully supported
- recruitment problem.
- No algorithm engineer before 2/15 deadline.
- No full-stack engineer
- no full-stack engineer right now, need help from someone from other team.
## Discussion
- printer issue.
- not be able to print
- missing some columns
- missing some terms in formula
- missing some figures
- recruitment problem.
- algorithm engineer: use connections from school
- engine research team adding workload onto us.
- how to deal with story that runs too long.
- wantwant

## Action item
- Remind PIC of action item via slack channel (Che-Fu)
- Monitoring if “wantwant” snack still exists(it grows!!) (Che-Fu)
- Ask if Lidar can be repaired (Che-Fu)
- Find out who's responsible for printer and make it fixed.(Cindy)
- Algorithm engineer
- Ask professor for potential candidate(Cindy)
- Ask friends (Hao-Yu)
- detailed JD for recruiting (Hao-Kai)
- Think about how to get full stack engineer(Hao-Kai)
- If Engine research team ask for help, just refer to Hao-Kai(Cindy, Hao-Yu, Hao-Kai)
- Ask for more details if a story runs too long(AID)
- Hand in WantWnat alternative list(Che-Fu)

# 20191223_retrospective

## Good
- Short spring is good for research work(+1)
- test procedure for visual quality is completed
- have used JIRA more actively

## Improvement
- Should send retrospective meeting invitation on time
- No design review--lack of documentation and reference review
- Still someone not updating daily scrum (maybe due to late or sick-leave)
- Some scope of sub-task is too large (4~5 work day)

## Impediment
- Lidar has been proven shitty :~~~
- Need one more CPL algorithm engineer (to paired with Jojo)
- Arial is not attending sprint planning
- Lab engine malfunction too often
- GPU is not available during night time
- JIRA is too noisy in slack channel
- PM is not understanding the content of daily scrum (ticket content is not related to the result of sprint planning)

## Discussion
- Rule #17, #18, saga

## Action item
- Discuss about the workflow of working with external team (Hao-Kai, Arial, Jojo, Cindy)
- Engineer will decide a deadline for technical requirment
- Make sprint slides as a story biweekly
- Discussion about Saga, direction (hao-Kai, che-fu)
- Remind PIC of action item via meeting invitation (Che-Fu)
- Make review of code to a story(AID)
- Monitoring if “wantwant” snack still exists (Hao-Yu)
- pre-planing meeting should make sure the main direction is clear (Hao-kai,Jojo, Arial)
- Should send retrospective meeting invitation on time (Che-Fu)
- Write design review, especially literature review (no matter if the experiment if successful or not) (CPL)
- Divide subtasks, even if the subtask is already in progress
- In sprint planning, write down the reason/purpose/PIC of the stories, and also write down the summary of the planning. (CPL/AID)
- Ask if Lidar can be repaired (Che-Fu)
- Find 1 more CPL Algorithm Engineer (Hao-Kai)
- Lab Engine related:
- Ask Stefan (and then Ethan if there is no deadline for our P0/P1 issues) how Lab Engine team maintains the infrastructure (Hao-Yu)
- Force engine team to review GPU quota (Hao-Yu)
- Use cpl-jira and aid-jira again, and delete chatbot in scrum channel (Che-Fu)
- Let rule 17 & 18 happen 1 more time and decide if we want to abandon them (Che-Fu)

# 20191206_retrospective
## Good
- many people are willing to help me, good, good, and so do I (+++)
- OKR is ready
## Improvement
- daily scrum is not fully fulfilled
- miscalculate the points needed for an issue(end2end)(++)
- winter comes, so do later comers.
- font in sprint slides is too small
- need to make review to a story
- no more overtime work
- lonely scrum
## Impediment
- too many "WANTS" snacks
- need some non-caffeine drinks(tea bag, or bottled tea, etc.)
- people should update daily scrum
- sprint slides should be ready before certain deadline
## Discussion
- When to finish sprint review slides?
- should we have penalty for late comers?
- how to remind everyone about easily violated rules?
- (PIC forget to update action item)
- When to come into the office? --- no need to discuss, not gonna help
## Action item
- Discuss about the workflow of working with engine team (Hao-Kai, Arial, Jojo, Cindy)
- Add JIRA stories directly in discussion meeting with Stefan
- Use Sprint goal to convince external team to agree with our job content (PM)
- prevent S3 appearing in CPL OKR.
- Create Sprint slides immediately after sprint planning(everyone)
- Make sprint slides biweekly
- OKR version control: use Google Doc "Name current version" button (everyone)
- Assign a version name to your current edition
- do not download and create another copy of file
- README should be written such that users can follow and will not misused the code (Hao-Yu)
- Make a new Master for sprint slides (Jojo)
- discussion about Saga, direction(hao-Kai, che-fu)
- Reason your argument during code review if there is conflicting, you dont just say "wrong" or "no way", be more patient (Hao-Kai)
- Ask Stefan if someone from engin team can review interns' code (Hao-Kai)
- update README while updating code, add to rule(Hsiao-Hsuan)
- add hyperlink of meeting guidelines in wiki(HH)
- Sprint slides should be ready the day before Sprint review(everyone)
- add to rule(HH)
- Remind PIC of action item via meeting invitation(Che-Fu)
- Make review of code to a story(AID)
- will request more accurate buffer of a particular work(Jojo)
- Ask for no "wantwant" snack and no-caffine drink(16-tea and flower tea bag)(Hao-Yu)
-
# 20191122_retrospective
## Good
- there is more research , no website!!
## Improvement
- unexpected event not planned {++}
- do A in B story
- if code is not general enough, it should not be misused
- story with deadline is not prioritized high enough
-
## Impediment
- Hsiao-Hsuan should learn emacs : (
- we don't have embedded system engineer
- endless code review (code architecture issue)
- something wrong with slides style for sprint review
- the goal/work item is not clear for garlic
- rule #17
- code review is frustrating [just a matter of coding style or real problem there]
- test case fail while model changed
- CPL player is set done but not usable
## Discussion
- how to prevent misused of code
- no one can review interns' code
## Action item
- Discuss about the workflow of working with engine team (Hao-Kai, Arial, Jojo, Cindy)
- Add JIRA stories directly in discussion meeting with Stefan
- Use Sprint goal to convince external team to agree with our job content (PM)
- prevent S3 appearing in CPL OKR.
- Design documentation for Coding {formula, references, etc} (AID)
- Create Sprint slides immediately after sprint planning(everyone)
- Make sprint slides biweekly
- OKR version control: use Google Doc "Name current version" button (everyone)
- Assign a version name to your current edition
- do not download and create another copy of file
- Ask Jojo about the link of current OKR (Hao-Kai)
- Remind Intern to put daily scrum content on slack channel (Hao-Yu)
- CPL should stick to the meeting of organizing sprint slides the day before sprint review day (CPL)
- README should be written such that users can follow and will not misused the code (Hao-Yu)
- update README while updating code, add to rule(Hsiao-Hsuan)
- should plan some buffer for unexpected events(AID)
- stick to story (Che-Fu, Hsiao-Hsuan)
- team should remind each other about job priority(CPL)
- one last chance of emacs (Hsiao-Hsuan)
- find resource for embeded system (Che-Fu)
- ask Stefan
- check context of code after code review modification to prevent inconsistency (everyone)
- update sprint review style to make it more clear (Jojo)
- discussion about Saga, direction(hao-Kai, che-fu)
- Reason your argument during code review if there is conflicting, you dont just say "wrong" or "no way", be more patient (Hao-Kai)
- Replace test case with visualization script (CPL)
- Ask Stefan if Chu-Hsuan can work on CPL player API (Hao-Yu)
- Ask Stefan if someone from engin team can review interns' code (Hao-Kai)