Re: [boost] Testing (was PLEASE READ: Another round of operators.hpp fixes)

From: Beman Dawes (beman_at_[hidden])
Date: 2000-07-05 06:49:28

Mark Rodgers wrote:

 >> My big problem with this test is that it doesn't tell me at a glance
 >whether
 >> it worked or not. It compiles and runs, but are the results correct?
 >
 >I have the same problem with the "test" for functional - it's really just 
a
 >compilation test and example rather than a test that the functionality
 >works.
 >
 >I think it would be nice if we could come up with a standard testing
 >framework so that all the testing could be uniform and the results 
readily
 >extractable by Beman's compilation script.  Dave, are you volunteering to
 >put together a standard we can all adhere to, and some helper code to
 >reuse? ;-)

There have been at least two suggestions that boost include a Unit Test 
library.  Presumably we would use it ourselves to test our other libraries.

My problem is 1) I'm tied up with other boost activities, and 2) don't know 
anything about unit test techniques.

So I would like to encourage others to try to come up with a boost testing 
(unit or otherwise) library.

Some Possible Requirements:

* Attractive enough to use that boost authors will want to use it.  To me 
that implies it has to be non-intrusive.

* Tests using it have to run automatically and yield a definitive pass/fail 
result. Etc.

* Use techniques known to be effective.  In other words, there is a huge 
software testing community out there, so build on their successes and don't 
reinvent the wheel.

The comp.software.testing FAQ at http://www.rstcorp.com/c.s.t.faq.html is a 
major resource with lots of links, bibliographies, definitions, and so 
forth.

--Beman