[Carpet] nPoints assertion

[Erik Schnetter]

On Mar 19, 2008, at 13:05:16, Ian Hinder wrote:
> Erik Schnetter wrote:
>> On Mar 17, 2008, at 16:45:42, Ian Hinder wrote:
>>
>>> Hi,
>>>
>>> There is an assertion in Recompose.cc (line 1137)
>>>
>>>     assert (mynpoints.at(n) > 0);
>>>
>>> which aborts my run when I run at low resolution on 16 processors.
>>> Erik, you suggested this check might not be necessary, and indeed  
>>> things
>>> work fine with the assertion removed.  OK to commit a patch or  
>>> should I
>>> do more specific testing?
>>
>>
>> OK to commit.  More specific testing would be difficult, since this
>> assertion failure will only catches illegal cases which are not  
>> supposed
>> to reach this point in code anyway.
>
> Now I'm confused.  If the cases are illegal how am I triggering the
> assertion?


I meant to say: I place assert statements into code to say "this is  
what the algorithm should be doing at this point".  If the assert is  
triggered, then either some previous piece of code was wrong, or my  
idea about the algorithm was wrong.  If everything works as planned,  
no assertion would be triggered.

In this case, my idea of the current state was wrong.  I assumed that  
each grid is so large that each processor can own at least one grid  
point.  This is clearly not always possible, therefore this assert  
statement should be relaxed (>=) or removed, since mynpoints is  
nonnegative anyway.

Testing it would require to find a case where this assert statement is  
triggered.  Since this should never be the case, it is very difficult  
to come up with a test case -- it would require writing custom code  
and calling this routine with known-to-be invalid arguments.

-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/20080319/afb2f3d6/attachment.pgp