Anonymous usage statistics

Parity and XamlContrast · 匿名使用統計

Two open-source developer tools, Parity (design fidelity checks) and XamlContrast (contrast audits for XAML apps), send a small set of numbers here after each run. This page says what they send, why, how long it is kept, and how to turn it off.

Parity(設計還原度檢查)與 XamlContrast(XAML App 的對比度稽核)每次執行後會送一小組數字到這裡。 這一頁說明送了什麼、做什麼用、保存多久、怎麼關掉。

Why · 為什麼

For academic research: how far real implementations drift from their designs, and what contrast problems real XAML apps have. It is supplementary evidence next to a fixed, reproducible set of projects — not a replacement for it, and not used for anything else (no advertising, no profiling, no sale). Results are published only in aggregate.

用於學術研究:實際的實作與設計稿差多少、真實的 XAML App 有哪些對比度問題。它是固定、可重現的專案語料 之外的補充證據,不取代它,也不做其他用途(不做廣告、不建個人檔案、不出售)。只公開彙總後的結果。

What is sent · 收什麼

Only counts and ranges. Every request is checked field by field, and anything with an extra, missing or out-of-range field is rejected and not stored.

Both toolstool and version, operating system family (windows / linux / macos / other), whether it ran in CI, and a project ID only if you put one in your config file yourself
Parityfidelity score (0–100), whether the gate failed, number of pages, design source type (figma / snapshot / file / image), design node and matched node totals, gap counts per severity, and per measured property (for example paddingTop or color) how many gaps and how large, in ranges (for example 2–4 px, or a colour difference of 2–5 ΔE)
XamlContrastthe audit verdict (pass / below the gate / not trustworthy), the gate and palette mode, file and colour-pair totals, coverage, fail / warn / ok / decorative counts, unresolved counts per reason, and how many pairs fall in each contrast range (below 1.5:1 … 7:1 and above)

只有計數與區間。每一筆都逐欄驗證,多一個欄位、少一個欄位或超出範圍就拒收、不存。 兩個工具都送:工具與版本、作業系統類別、是否在 CI、以及只有你自己寫進設定檔時才有的專案 ID。

What is never sent · 一律不收

URLs, routes, page or element names, CSS selectors, file names or paths, screenshots, colour values, design content, tokens, repository names or IDs, user or machine names. The tools send no User-Agent and no cookies.

IP addresses: Cloudflare, which runs this service, sees the connecting address like any web server's network provider does. This collector uses it only in memory to rate-limit, and never writes it to the database or a log. Only the UTC date of each submission is kept — not the time.

一律不收:網址、route、頁面或元素名稱、selector、檔名或路徑、截圖、顏色值、設計稿內容、token、repo 名稱或 ID、 使用者或機器名稱。工具不送 User-Agent、不送 cookie。
IP:代管這個服務的 Cloudflare 跟任何網站的網路服務商一樣看得到連線位址;收集站只在記憶體裡用它限速, 不寫進資料庫、不寫進 log。每一筆只記 UTC 日期,不記時間。

Turning it off · 怎麼關

Any one of these stops it:

To see exactly what a run would send without sending anything, add --stats-preview. A failed submission never changes a tool's result or exit code; it waits at most 3 seconds.

任一個就不送。想看某次會送出什麼而不真的送,加 --stats-preview。送不出去絕不影響工具的結果或結束碼,最多等 3 秒。

Keeping and deleting · 保存與刪除

Kept until: 2030-12-31.

Without a project ID nothing links a submission to a person, a machine or a repository, so individual submissions cannot be found again — there is nothing to look up and delete. If you set a project ID and want its submissions removed, send that ID to: stats@cornhsu.com.

保存到:2030-12-31。沒有專案 ID 的資料對應不到任何人、機器或 repo,所以無法、也不需要個別找出來刪除。 設過專案 ID 而想刪除,把那個 ID 寄到:stats@cornhsu.com。