Ken Smith
On Fri, Nov 21, 2008 at 11:09 AM, Gennadiy Rozental
wrote:
Ken Smith wrote:
On Thu, Nov 20, 2008 at 11:47 PM, Moritz Beber
wrote: Hello everyone, after prying for any hints in the documentation and reading frustrating mails in the archives for a while now, I get the feeling that this is a very confusing topic. My aim is to use a static library of boost.test because I need to compile it a lot and the waiting time when using the header is just frustrating.
I felt the same way until I switched to using the minimal test suite.
I would be interested to know why.
The builds simply took much longer with boost/test/included/unit_test.hpp and I wanted to do the least amount of fiddling with my build to use this package.
How much longer it takes? In my experience on modern PC/Linux box the difference almost negligible. How difficult would it be to spend 10 min once and build the static library? In that case compilation time should b even better than minimal.hpp. IMO advantages way overweight, even if you do not need any advanced features immediately. Give it a shoot - you may find it useful.
I'm sad to hear the minimal test framework is not recommended as it was working just fine for me so far.
You are free to use it. It's not going anywhere. But from almost any standpoint you may find UTF better
Basically, I wanted to figure out how to do the least amount of work to implement unit testing so my colleages would push back as little as possible when I recommended they use it. More testing is better testing regardless of the mechanism.
With UTF you can get more from the testing. Gennadiy