Skip to Content (Press Enter)
Community › Forge › Assets list › ODC Native Chunking Tester
Login to follow
ODC Native Chunking Tester

ODC Native Chunking Tester (ODC)

Stable version 0.1.1 (Compatible with ODC)
Uploaded on 12 Jun (11 days ago) by DB Results Labs
ODC Native Chunking Tester

ODC Native Chunking Tester (ODC)

Details
Detailed Description

An administrative preview dashboard and testing harness designed to audit and benchmark the native Semantic Search text-splitting configurations in OutSystems Developer Cloud (ODC).

Because ODC's ingestion pipeline and PostgreSQL pgvector index are fully encapsulated inside platform-managed background cloud infrastructure , developers face a "black box" regarding how their data is structurally partitioned.

This component lets you safely stress-test your ingestion strategy against high-volume test payloads (TestFile.md and TestFile.txt) using the SaveTestContent timer. It simulates all 4 native strategy behaviors (Smart, Fixed-Size, Sentence-Based, and Recursive) directly on an interactive UI. This enables you to inspect character splits, block sizes, and matching boundaries before committing data to a live vector index.


Read the deep-dive research article that inspired this component: What ODC Actually Does When It Chunks Your Text

ODC Asset screenshot_0ODC Asset screenshot_1ODC Asset screenshot_2ODC Asset screenshot_3ODC Asset screenshot_4
Limitations

  • ODC Environment Dependency: The component is entirely dependent on native Semantic Search being actively provisioned and enabled in your ODC tenant instance to function.

  • Fixed Data Set Payload: The testing canvas is scoped to evaluate the pre-compiled sample resources (TestFile.md and TestFile.txt). It does not support custom ad-hoc documentation uploads directly via the UI interface.

  • Static Parameter Bounds: Simulation runs use standard ODC defaults (1000 character limits, 200 overlaps, and 1 sentence limits). It does not dynamically modify live database configuration tables out-of-the-box.

  • Asynchronous Index Latency: The component does not display results instantly; you must wait briefly for ODC’s asynchronous background cloud AI layer to finalize text processing after running the seeding timer.

  • Non-Production Testing Target: Designed purely as a design-time simulation and audit utility; it is not built to act as a production data pipeline or transformation tool.

Release notes 

Added page links on the menu