Re: [pooma-dev] RFA: Patch for fixes for regression test failures

[Jeffrey Oldham]

On Tue, Oct 09, 2001 at 06:35:25PM -0600, Mark P Mitchell wrote:
> 
> The attached patch attempts to fix several regression tests.
> 
> Here is a summary:
> 
>   src/Array/tests/ScalarAdvection.cpp:
>  
>     Updated to use approximately the same code that Scott used in 
>     examples/Field/ScalarAdvection/ScalarAdvection.cpp.  
> 
>     Note that although this test now compiles, it still fails for dimension 
> 2.  Scott's code fails in dimension 2, as well.  This code works in dimension 
> 1 as does Scott's code.  

Was the original code dimension-independent?  If so, it would be nice
that the revised code is also dimension independent.

>   src/Connect/Lux/tests/*:
> 
>     These tests failed because LuxConnector.Field.h had not been 
>     updated for the new fields.  

For me, lux_test2 still fails, but I believe that is a particle problem.

>   src/DynamicArray/tests/dynamic_array_badcreate.cpp:
> 
>     This test tests to make sure that you cannot dynamically change the
>     size of a DynamicArray when there is a view outstanding for the
>     DynamicArray.  (Doing so could leave the view pointing at 
>     nonsense.)
> 
>     However, nothing in the DynamicArray code actually seems to
>     enforce this constraint, so the test fails at run time.
> 
>     I couldn't fix this one.  Should the test be changed?  Should
>     such a check be implemented?
> 
>   src/Field/tests/Positions.cpp:
>   src/Field/tests/CrossBox.cpp: 
> 
>     Also needed updating for the new fields.
> 
>   src/Pooma/tests/pooma.cpp:
> 
>     This test was written wrong -- it was not set up to exit with
>     code zero on success.
> 
> OK to check in?

Yes.  I have attached a slightly different patch.  Would you please
list on which platforms you tested the code?  I compiled and tested
the code on Linux/g++.

Thanks,
Jeffrey D. Oldham
address@hidden

foo.patch
Description: Text document