Dimi Paun wrote:
On Mon, 2005-11-14 at 03:32 -0600, Alexandre Julliard wrote:
The only way to know that something is done is to write a test case for it and check the behavior against Windows. MSDN cannot be trusted, and marking code as DONE just because it follows what happens to be in the doc today is very misleading.
I'm not arguing against test cases -- it's obviously better to have test cases. However, even if we do have them, there's no way of knowing that we are testing all documented aspects. In fact, I doubt we have such exhaustive tests right now.
I know that the docs are not perfect, but they are still an important milestone, and a relevant one in the vast majority of cases. It's been important (for me at least) to know the level of completeness versus documented behavior for the common controls. This is useful today, and it's a fraction of the effort compared to writing tests. I think they are complementary.
and how do you deal of MSDN docs evolving ? I'm not even speaking of evolution because of new features, but MSDN gets updated from questions asked on the KB. A+