Re: [platform-core-dev] Testability experiment

[Magiel Bruntink <magiel@xxxxxxxxxxxxxxxxxxxxxx>]

On Thursday 15 May 2003 16:40, John Arthorne wrote:
> I guess it depends on what you mean by a unit.  We certainly test units of
> API functionality... For almost every API interface there is a
> corresponding test class that has the same name as the API interface with
> the suffix "Test".  For example IFileTest is a unit test for IFile,
> IResourceTest for IResource, etc.  We then have non-API tests for
> significant sub-systems of the resource model, such as builder, the
> property store, data tree, etc.  These tests don't operate on individual
> classes, but groups of related classes.  I think to assume that the unit
> of testability always maps to an individual class is an unneccesary and
> unrealistic assumption.  Just my two cents...

Certainly it depends on the definition of unit. I'm not really limiting myself
to investigating testability at the class level, although it has been my primary
focus for several practical reasons. Apparently the most appropriate
definition of `unit' depends heavily upon the test process. The test suite
of the project that is going on at the company I work for is very class-oriented.
Hence also my initial focus on the class level.

You say that the Eclipse test suite tests groups of related classes at a time.
Should I read `package' for these groups of related classes, or a different
unit still?

Yours,
Magiel