Difference between revisions of "Category:Testing"

Revision as of 09:22, 30 May 2007

A systematic testing process is necessary to produce a timely release with a high level of quality. The last six weeks of a release cycle are entirely devoted to testing. The general organization is described below.

At this level of the development all testing is performed by a team member other than the developer of the corresponding system or subsystem. The testing process is formalized:

• For each system and subsystem, a Test Plan is created accordingly.
• The test plan is made of a number of descriptions of tests to be run, each known as a Test Case Specifications (TCS). A TCS specifies a test scenario; it is not related to any particular execution of that scenario.
• The testing process consists of running these test scenarios; the result of any such run is a Test Run Report (TRR).

For the compiler, we use the Eweasel regression testing tool. It consists of a set of systems that are compiled. The tool checks for each system that an error is reported if one should, or that no error occurs if none are expected. It also checks that executing a system yield the right execution semantics.