[Insight-developers] Empty FixedArray destructor: Performance hit using gcc (times 2) : What criteria to make the test fail ?

[Luis Ibanez]

Bill,

Thanks,

It was missing #include <time.h>
I just committed a fix.


---------

BTW:

It is still unclear what the "failure" state should be for this test.

Two options that come to mind are:

    A) Make it fail if the alignment is 4-bytes

    B) Make it fail if the time for running the
       loop between the 4-byte alignment and 8-byte alignment
       changes by more that x%, maybe 50% ?


Any suggestions ?


    Luis


--------------------
Bill Lorensen wrote:
> Luis,
> 
> That test is not compiling on Borland.
> http://www.cdash.org/CDash/viewBuildError.php?buildid=93460
> 
> Bill
> 
> On Fri, Jun 6, 2008 at 11:03 AM, Luis Ibanez <luis.ibanez at kitware.com> wrote:
> 
>>
>>Hi Tom,
>>
>>
>>I have no objection to removing the empty destructor of the FixedArray.
>>
>>
>>However, removing that code without understanding "why" is that it makes
>>a difference for the alignment, seems to be a dangerous way to maintain
>>the toolkit.
>>
>>
>>We are also badly place at this point to judge on how much this issue
>>matters for other platforms.
>>
>>
>>We will know more about it tomorrow morning when we see the outcome of
>>itkFixedArrayTest2 in the Dashboard.
>>
>>
>>In either case, we must document this observations in the code,
>>to make sure that the effect of the destructor presence (or absence)
>>is not forgotten.
>>
>>
>>
>>   Luis
>>
>>
>>------------------------
>>Tom Vercauteren wrote:
>>
>>>Well I would really vote for removing the empty destructor
>>>implementation. This seems to be the cleanest way to solve the
>>>problem. The empty destructor is useless and only interferes here. I
>>>don't really like changing compilation flags or adding macros and
>>>adding a new derived class would be confusing for the user.
>>>
>>>Tom
>>>
>>
>>_______________________________________________
>>Insight-developers mailing list
>>Insight-developers at itk.org
>>http://www.itk.org/mailman/listinfo/insight-developers
>>
> 
>