Hi All,

Given the facts; that code coverage data is most relevant in white box testing (Unit and Component Testing) and that the in-house Symbian Foundation smoke and sanity testing is on a system test level, then code coverage data is best captured when packages are submitted to the Foundation.

The code coverage data is not changing much after Function Complete and not of great value measuring during hardening phase. It should be recommended/enforced that package submissions are delivered with test results and the associated code coverage. The next step will be to set a quality standard value for Functional and Structural coverage.

Looking some other open source community projects the practise seems to be that a "quality council" runs tests and measure code coverage during a quality build-up phase. The "quality council" will be made-up of different interested foundation members. One possible implementation of this approach could be that package owners on a regular basis submits test results and code coverage data to be published by the Foundation.

--Sf 17:06, 21 August 2009 (BST)

Hi Stefan,

I agree with your point of view. I don't think that we should or can enforce code coverage, but it should be consider best practice.

Perhaps we should move way from demanding the use of an specific tool and focus more on measuring the availability of data. To ensure that the data is reliable it makes sense for symbian staff to evaluate tools that package owners are using and ensure that they are fit for propose.

In summary, I would like to recommend that we ask PkOs to provide code coverage result for their packages and state what tool they used to collect them. This metric should focus on reporting what packages have code coverage figures.

any thoughts?

--Victor 13:32, 3 September 2009 (BST)