Skip to content

regression_task.md

Latest commit

 

History

History
49 lines (32 loc) · 1.48 KB

regression_task.md

File metadata and controls

49 lines (32 loc) · 1.48 KB

Regression Task

The regression_task identifies the commit range where a regression was introduced. It uses binary search to narrow down the range of revisions that caused the crash.

Key Responsibilities

  1. Build Setup:
    Sets up builds for specific revisions to test for the crash.

  2. Binary Search:
    Uses binary search to efficiently identify the regression range.

  3. Validation:
    Ensures that the identified regression range is accurate by testing earlier revisions.

  4. Task Creation:
    Creates follow-up tasks, such as blame analysis, based on the regression range.

Workflow

  1. Initialization:

    • Fetches the test case and crash information.
    • Sets up the test case and its dependencies.

  2. Build Setup:

    • Sets up builds for the revisions to be tested.

  3. Binary Search:

    • Tests the crash at the middle revision of the current range.
    • Narrows the range based on whether the crash reproduces.

  4. Validation:

    • Tests earlier revisions to ensure the regression range is accurate.

  5. Finalization:

    • Saves the identified regression range.
    • Creates follow-up tasks for further analysis.

Important Notes

  • Bad Builds:
    If a build is invalid, the task skips it and continues testing.

  • Validation:
    The task ensures that the regression range is accurate by testing additional revisions.

  • Error Handling:
    If the task cannot complete due to errors or timeouts, it retries later.