Test coverage

Looking at the test coverage results, I think we are getting lower numbers
than might be expected as assertions show up as branches not taken. By
their nature, most of those assertions should be impossible branches! So
should we be running the test-coverage tests with assertions disabled?

If this is already the case then I'm going to have to learn to read the
output better!

AlisdairM

Hi Alisdair,

Yes, you are right its probably worth running on a -Asserts build for
this. For now I think the coverage results are mostly useful for
finding uncovered *lines* of code, I haven't tried to optimize the
process for better numbers since there is still plenty of work to do
there.

- Daniel