[Insight-developers] nrrd assumptions and signed overflow

[Sean McBride]

They have fixed this upstream, but as I don't have time to learn how to properly merge with upsteam, I've created this trivial patch:

<http://review.source.kitware.com/3721>

Which should fix many of the failing tests on our (Rogue) dashboards.

Sean



On Mon, 28 Nov 2011 16:39:15 +0000, Williams, Norman K said:

>Well, I guess that means CLang agrees with me that the NRRD code is just
>plain wrong.
>
>I'd vote for a patch to NRRD in ITK that removes the code that
>intentionally overflows.
>
>
>
>On 11/28/11 9:27 AM, "Sean McBride" <sean at rogue-research.com> wrote:
>
>>On Mon, 28 Nov 2011 14:50:53 +0000, Williams, Norman K said:
>>
>>>I presume CLang is testing for known overflows at compile time.  Does
>>>this
>>>get flagged as a compile error?
>>
>>No, it's a check at runtime.  It's very nicely explained here (well worth
>>the read):
>>
>><http://blog.llvm.org/2011/05/what-every-c-programmer-should-know.html>
>>
>>All the "ILLEGAL" type failures here:
>>
>><http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=1768103>
>>
>>are due to these checks.