Java analysis is 30% faster on average

Hello Java developers,

We did significant changes in our Java analyzer so it runs faster. Our tests across different project sizes show a reduction of 30% of the execution time on average. One beta tester said analyzing their 1 million LoC project dropped from 38 minutes to 18.

We believe we reached a limit and it would be hard to get an even faster Java analysis time while scanning all the files of a project. This is why our next objective is now to analyze only changed Java files on pull requests.

Enjoy!
Alex

8 Likes

Is there any (rough) timeline on when the next objective (only analyze changed Java files) will be reached? We would love this feature (and it would allow us to analye our whole project, which means more billing for you if this is any incentive…)

Anyways, I am already glad you improved the Java performance. We will monitor our runtimes and see how it affects us.

2 Likes

@dirkkirsten We don’t need an incentive, we are already super motivated and excited to deliver “only analyze changed Java files”. Regarding the timeline: Q2 for SonarQube, probably Q3 for SonarCloud.

2 Likes

@Alexandre_Gigleux Is there anything special we need to do or can do to use the new analyzer version particularly for a SonarCloud project that does analysis via our CI provider?

1 Like

@Rick-at-KUBRA You have nothing to do to benefit from this change. If you don’t see any difference, I suggest that you create a dedicated thread so we can have a look (please share some details about your build: OS, CPU, RAM, memory allocated to the scan, LOCs of your project)

1 Like

A post was split to a new topic: Analysis performance decreased due to “javabugs” Sensor