Skip to content

[Performance]: buildchecks #11216

Open
Open
@SimaTian

Description

@SimaTian

Issue Description

After some initial benchmarking, the buildchecks have ~8% performance cost.

Steps to Reproduce

Run build checks on an orchardCore solution, use maxcpucount:10 on a dev drive, use a cold run.

Data

initial and rather crude benchmarking, however the impact is visible.
Left column is a baseline, right one is using build checks. Last column are averages.
47.04163 |   | 57.319
49.99453 |   | 52.50005
49.50645 |   | 52.69354
52.76786 |   | 50.99738
47.06608 |   | 50.97427
47.69965 |   | 55.23598
  |   |  
49.0127 |   | 53.2867

Analysis

While the benchmark I used is basic and rather crude, the result hint at a regression in a ballpark of 6-10%.
This is to be expected however it poses an issue if we want to make build checks opt-out.

One possible solution could be to implement caching - the project files usually don't change all that often so only re-running build checks when an update happens should be reasonable.

Versions & Configurations

No response

Regression

  • yes
  • no

Regression Details

No response

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions