[Rivet] Bug: namespace and include pollution in Rivet namespace (minor)

[Andy Buckley]

On 30/09/14 09:47, David Grellscheid wrote:
> 
>>> 2. assert() can be used without "#include <cassert>"
>>
>> Is that a bad thing? We include it in Rivet/Tools/RivetSTL.hh, along
>> with the various "using std::*". User convenience again. The header
>> chain needs to be cleaned up (any good tools to help with disentangling
>> #include messes?) but that's a separate issue.
> 
> Here I agree with David. Since assert() is a macro controlled at compile
> time by NDEBUG, it cannot be safely used in header files anyway. Every
> .cc that needs must include it locally itself.

Ah, I see. I've come to the conclusion that assert is a real mess
precisely because of that (maybe inconsistent) compile time
en/disabling. And hard fails are rarely nice: I've been replacing
asserts with exceptions in many places in YODA.

Anyway, it's something to fix but should not be a release blocker. I
checked and we have quite a lot of uses of assert() in headers -- more
than a quick fix, certainly. Perhaps we should provide our own
definition of it (or a RIVET_ASSERT), to bypass the NDEBUG issue... just
a thought.

Andy

-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN