[lsb-discuss] ALSA testing != 0

Stew Benedict stewb at linux-foundation.org
Wed Apr 29 04:39:38 PDT 2009


Denis Silakov wrote:
> Robert Schweikert wrote:
>   
>> During the F2F we bumped against the test coverage concern w.r.t. ALSA a
>> few times. While the coverage is not stellar, it is also not as bad as
>> we thought as the attached analysis shows.
>>
>> I am not sure how we track upstream testing within our coverage data.
>> Further I made the assumption that the tests are valuable, i.e. more
>> than "the interface exists and doesn't crash".
>>     
>
> Well, we don't track upstream coverage itself; we only track coverage
> for tests included in the certification suite.
>
> A usual process is to adopt existing upstream tests (if any) in LSB test
> suites - drop tests for non-LSB symbols, make the tests themselves LSB
> compliant, etc..
>
> (I guess Stew can say more about integration of upstream tests into LSB
> test suite).
>
> In general, it seems that we should investigate existing libasound tests
> and try to use them.
>
>   
My usual process is:

1) See how the tests can be packaged/run in a standalone fashion, 
against the system library rather than the build tree (for typical unit 
tests)
2) See how I can patch in support for tet journal generation. Sometimes 
it's easy, if the test harness is adaptable.
3) Try to build the tests with lsbcc, patching as needed

We can certainly open a bug with the alsa test info, so it's captured 
somewhere.

-- 
Stew Benedict
Linux Foundation




More information about the lsb-discuss mailing list