[v3] Refined test instrument for container tests

[Stephen M. Webb]

On January 9, 2002 03:08 pm, Benjamin Kosnik wrote:
> > Um, in particular the ones sitting on my hard drive undergoing
> > refinement.  I've slowly been working on a bunch of 23_containers
> > tests like the list_* ones.
>
> Ok well that's great news, thanks. Are you thinking of exception-type
> tests a la STLPort?

Um, I'm not sure since I didn't know STLPort did exception testing.  I 
am lery of looking at other code too closely in case I get accused of 
plaguarism.  On the other hand, one can see farther standing on the 
shoulders of giants.  Maybe I'll just cd over there and take a look...

I've been basing my tests on the standard.  If it explicitly mentions 
exception effects or complexity bounds (eg. the copy constructor of T 
is called at most N times) and I can figure out a way to reliably test 
it, I write a test case.  Sometimes the standard is, well, vague or 
inscrutable, so coverage is not as complete as I'd like.

I have found that on occasion tests based on my interpretation of the 
standard fail with libstdc++-v3.  I am assuming my interpretation is 
wrong -- wouldn't be the first time -- but I wouldn't mind the 
occasional expert ruling on corner cases.  What would be the best way 
to get an interpretive ruling?  Post to comp.std.c++ and hope to get a 
reply?

> I guess right now just proceed as you are, and we can figure out a
> way of organizing memory-limits/container-checking/benchmarking
> later.

Will do.

__
Stephen M. Webb