[Carpet] status of CarpetSlab

[Erik Schnetter]

On Mar 13, 2008, at 06:07:32, Thomas Radke wrote:

> Erik Schnetter wrote:
>>> I tried to rewrite this call to something similar in CarpetIOASCII  
>>> but
>>> then got stuck when entering singlemap mode - in which mode should
>>> Hyperslab_Get be called ?
>>
>>
>> It should be called in singlemap mode.
>
> But then the code in CarpetTest/src/slabtest.cc:48 must be wrong: a  
> grid
> function pointer 'yy' is passed as hdata argument into Hyperslab_Get()
> but this pointer is always NULL in modes other than LOCAL.

For backwards compatibility, if there is only a single patch, and if  
there is only a single component per processor, then it should also  
work to call CarpetSlab in local mode.  (This may break if there are  
multiple components per processor, since the routine which calls  
CarpetSlab is then executed multiple times.)

>>> It seems that CarpetSlab has been broken for a long time already (at
>>> least since December 2006 when the nightly Cactus integration  
>>> tests  had
>>> started). It is not being used anywhere except in CarpetTest.  
>>> Should  it
>>> be moved to CarpetDev ?
>>
>> Yes, I we can move it to CarpetDev, or maybe CarpetAttic since we  
>> have
>> no concrete plans to improve it.
>
> And CarpetTest too ?


CarpetTest tests also other things, such as e.g. the array layout in  
memory.  I would just disable the slab test routine instead.  The test  
case output does currently not depend on the hyperslabbing results.

-erik

-- 
Erik Schnetter <schnetter at cct.lsu.edu>   http://www.cct.lsu.edu/~eschnett/

My email is as private as my paper mail.  I therefore support encrypting
and signing email messages.  Get my PGP key from www.keyserver.net.



-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 194 bytes
Desc: This is a digitally signed message part
Url : /archives/developers/attachments/20080313/43c4ecd7/attachment.pgp