Re: [boost] Better testing; was type_traits_test

From: Beman Dawes (beman_at_[hidden])
Date: 2000-11-02 19:17:51

David Abrahams wrote:

 >> I'm starting to modify the regression test driver program to execute as
 >> well as compile the test programs...
 >
 >It all sounds good, but it won't fix the particular problem I'm having,
 >which is that the tests clearly need to be modified (perhaps a simple
 >#ifdef can be removed) in order to exercise more of the code under MSVC.

I know.  But maybe actually executing the tests regularly will encourage 
people to write tests with better coverage, and then maintain them.

Once we get a bit further, we can write a boost test policy describing what 
we expect.

Here is a related question.  For console applications such as tests, I use 
an #included main() function that acts as an exception catching wrapper for 
my real main() function (actually named xmain()).  This has the effect of 
(1) displaying what() and similar messages, and (2) turning exceptions into 
just another "return EXIT_FAILURE" for easy detection by test scripts.

See attached for a simplified version of the main() wrapper I use.

Would this be a good idea for boost testing?

--Beman