CUnit testing

[Björn Fahller]

On Friday 29 October 2010 03.13.35 Greg Banks wrote:

> Fascinating...it's always nice to see yet another test infrastructure
> implementation.

It is, isn't it? There's seemingly an endless need for small variations.
>
> I've only scanned the User's Guide, so please correct me if I'm wrong,
> but I don't see any significant advantage over CUnit + Valgrind?

Depending on what you want to test, it may or it may not. If you want to test 
abnormal situations, it does indeed offer a great advantage. Seeing that this 
is  not your focus, though, it is of little importance.

> And it
> requires C++, which nothing in Cyrus uses so far (and hopefully nothing
> ever will).

This, however, is a bit of a misunderstanding. Yes, the tests are written in 
C++, but you use it to test anything under the sky that can be called from 
C++, which is typically C or C++, but it may as well be Fortran or Ruby.

> Also, rather conveniently CUnit is already packaged on
> Ubunbtu where I do my development.

Obviously a convenient advantage.


> I'm quite a way away from testing abnormal conditions, although that
> will be very useful when testing the mailbox code.

Then, you will not find any advantage worth mentioning, until you get there.

   _
/Bjorn