[Users] fix WeylScal4::teukolskyID testsuite

[Ian Hinder]

On 22 Jun 2010, at 17:50, Ian Hinder wrote:

>
> On 22 Jun 2010, at 16:58, Roland Haas wrote:
>
>> Hello all,
>>
>> Ian Hinder found that the testsuite teukolskyID for WeylScal4 was
>> poisoned. It turns out I had forgotten to add MoL to the list of
>> ActiveThorns in the testsuite.
>>
>> Attached please find a patch which rectifies that. Ok to apply (both
>> development and release branch)?

I would be OK with that, on the basis that it is only the test which  
is affected, and the test wasn't meaningful before.  (We really need  
to have some sort of formal verification tests in addition to  
regression tests.)

>> Expect some more patches while Tanja and I try and add some extra
>> statements to WeylScal4 to make sure MoL is present (ie. something
>> like
>> inherit from MoL maybe) and adapt to a possible future change in how
>> MoL_PseudoEvolution is scheduled.
>>
>> Ian also pointed out some issues with how (outer hopefully)  
>> boundaries
>> are handled which will have to be looked at eventually.
>
>
> Since I am currently working on enhancing Kranc to automatically
> generate the code to correctly do the right thing with the boundaries,
> this should make the current "generate then patch" strategy obsolete.
> As of now, with my enhancements to Kranc, I can generate a version of
> WeylScal4 which passes the testsuite.  For the released branch, I
> suggest we just leave it as is.  For the trunk, I think we will want
> to autogenerate the code correctly, so I wouldn't spend time on fixing
> up the patch.


The attached patch to WeylScal4 (just the m directory; I haven't  
included the autogenerated files in the patch), to be used with the  
corresponding patches I sent to the Kranc list, removes the post- 
processing that was needed before to get the boundary treatment  
correct.  I also added "methodoflines" to the list of inherited  
implementations, since the thorn relies on MoL_PseudoEvolution being  
present.  Probably Kranc should make all thorns inherit from MoL,  
which would avoid this problem in future.

With the patches to Kranc and WeylScal4, WeylScal4 passes its existing  
testsuite (you have to delete a couple of now-unused parameters from  
the testsuite parameter file).  Depending on comments, I would propose  
committing these changes to the trunk.  There's no need to modify the  
release branch.

-- 
Ian Hinder
ian.hinder at aei.mpg.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: WeylScal4-AutoKrancBdry.patch
Type: application/octet-stream
Size: 7057 bytes
Desc: not available
Url : http://lists.einsteintoolkit.org/pipermail/users/attachments/20100622/a9337485/attachment.obj