Chinh (lelouvincx) / 2025-10-01

Created Wed, 01 Oct 2025 00:00:00 +0000 Modified Mon, 25 May 2026 06:02:25 +0000
403 Words

Note

  • Cross-model calculation

    • How we explained it?
      • Show the documentation.
      • Briefly describe the demo ecommerce dataset.
      • Explain the cross-model calculation as a metric in a dataset.
        • A calculation which takes 2 fields from different models, then multiply them together.
      • metric order_value {
        	definition: @aql order_items.quantity * products.price ;;
        }
        
      • Compare with their traditional approach (Holistics 2.7, 3.0, LookML, etc).
    • How customers perceive the value?
      • Timo (already use Holistics, migrating from 3.0 to 4.0)
        • Quy hết về 1 chỗ thì dễ kéo thả. Thay vì phải vào từng model để tạo measure.

        • Làm vậy sẽ giảm được số lượng các model trung gian phải tạo ra riêng cho từng widget.

      • Oddle (Syed) (already use Holistics, migrating from 2.7 to 4.0)
        • Syed thinked at model-level (since they were 2.7): “one model can refer another model”

        • Personal thought: Syed seems didn’t understand the strength of cross-model calculation (our semantic layer) => so I’m pretty sure the their migration approach is to replicate all SQL queries to every widget.
      • Aarki (Marshall) (known Looker)
        • In LookML, to perform a calculation between 2 fields from 2 different tables, you have to add a derived SQL query that do that particular calculation beforehand. And then you include that derived table into your dataset. And strangely your dataset has 2 tables that do the similar thing.

        • This is something more than I expected.

        • Reduce the amount of derived SQL to create.

        • Marshall told his experience at last company. We’ve been spinning up many “explorers” for slightly different data so that they could get a slightly different measures => and so this was so handy.

        • Can use this along with aggregation awareness to optimize query performance.

    • How would you improve?
      • Goal: customers may have various backgrounds (either migrating, or using another BI tool), the final goal is to make them feel Holistics’ semantic layer is strong and flexible.
      • Solution approach: solve their specific use case by AQL.
    • Minor fix:
      • During the call with Aarki, Huy got trouble showing Marshall the relationship => either show the relationship in docs or a better & faster way to show in demo4 to make customers get a quick grasp about the demo dataset.
  • Done

    • DONE Support Oddle listing all entities in their tenant - version 2.7

      • Clarify with Wudi.
      • Provide a Dashboards list.
      • Send message.
      • Wait for Syed’s response.
      • Side note to him: Data Imports + Data Transforms + Email Schedules.
    • DONE Review cross-metric calculation documents