Dasbor

Tempat pertanyaan yang disematkan dari workbench tinggal: setiap dasbor adalah kumpulan tile tersimpan, dan setiap tile menjalankan ulang query-nya setiap kali dibuka, jadi angkanya hidup, bukan potret. Pustaka mencari berdasarkan nama; papan dibuat, diganti nama, digandakan dan dihapus.

The library

Every board in the project, searchable by name, with the count in the page head describing the whole library rather than the narrowed view. New dashboard names a board and opens it empty; a board's row menu renames, duplicates or deletes it, and deleting asks first because a board's tiles are the only copy of their saved questions. How the library presents, as cards, rows or a table, is a stored preference per person.

A board and its tiles

Each tile is one saved question: its title, the current total, a signed delta against the preceding window and a sparkline of the series. Tiles are added by asking a question in the Analytics workbench and pinning the result here; the pin dialog offers every existing board and a new one. A trend tile carries the whole query forward: every metric, any formula, its breakdowns and its compared period, so a re-run always answers exactly what was pinned. A tile that fails to load says so on the tile, with its own retry, rather than poisoning the board; Refresh re-runs every tile at once, and a tile's menu renames it, switches it between standard and wide widths, moves it along the grid, or removes it.

Where this meets the API

The concept behind the boards is the Analytics pillar. On the wire: the library is the dashboards query, boards change through createDashboard, updateDashboard and deleteDashboard (pinning a tile is an updateDashboard carrying the new tile), and every tile renders by running its stored query through runQuery (GraphQL reference).