s4member environment and 'useless' tests

[Andrew Bartlett]

On Wed, 2017-01-11 at 09:17 +0000, Rowland Penny wrote:
> 
> It is understood why it is failing, it is failing because the test is
> against an unused entity.

No, it failed because of a serious use-after-free bug that metze just
fixed.  This would have been a show-stopper for 4.6 if not resolved. 

> 
> Can I ask a question, what if you have code that is 100% perfect but
> a
> test fails, which would you fix ?
> I get the feeling that you would fix the code. LOL

Every test tells us something.  I've not yet met code that is 100%
perfect, but a test that changes when the code change wasn't expected
to change the test output is always worthy of investigation.

Andrew Bartlett