[Insight-developers] Dashboard status & CMake version

[William A. Hoffman]

Moving to 2.0 would not be a bad idea.  It is the default download for cmake,
and has been out for about 8 months now, and it makes quite a few things easier
to do.   The move from 1.8 to 2.0 should be easier than 1.6 to 1.8 was.   

-Bill


At 01:40 PM 1/23/2005, Bill Lorensen wrote:
>In the meantime, I'll build cmake 2.0 on all of my unix/linux platforms and see what the impact will be.
>
>Bill
>
>At 12:26 PM 1/23/2005, Luis Ibanez wrote:
>>Bill,
>>
>>Well, maybe we could explore other options for getting around
>>this issue.
>>
>>Here is the relevant code:
>>
>>
>>Insight/Utilities/gdcm/src/gdcmCommon.h, lines 62-67
>>
>>> #if defined(ITK_SIZEOF_LONG_LONG)
>>> typedef  unsigned long long uint64_t;
>>> #endif
>>> #if defined( ITK_SIZEOF___INT64 ) && !defined( ITK_SIZEOF_LONG_LONG )
>>> typedef  unsigned __int64 uint64_t;
>>> #endif
>>
>>
>>and
>>
>>Insight/Utilities/gdcm/src/gdcmUtil.h, lines 666-696
>>
>>>#if defined( ITK_SIZEOF___INT64 ) || defined( ITK_SIZEOF_LONG_LONG )
>>>   unsigned char addr[6];
>>>   uint64_t n = 0;
>>>   int stat = GetMacAddrSys(addr);
>>>   if (stat == 0)
>>>   {
>>>      // Horner evaluation
>>>      for(int i=0; i<6; i++)
>>>      {
>>>         n *= 256;
>>>         n += addr[i];
>>>      }
>>>      // we fit on 15 bytes maximum < 256^6.
>>>#if defined(_MSC_VER) || defined(__BORLANDC__)
>>>      return Format("%I64u", n);
>>>#else
>>>      return Format("%llu", n);
>>>#endif
>>>   }
>>>   else
>>>   {
>>>      dbg.Verbose(0, "Problem in finding the MAC Address");
>>>      return "";
>>>   }
>>>#else
>>>  dbg.Verbose(0, "Problem in finding the MAC Address. This platform doesn't support 64bit integers.");
>>>  return "";
>>>#endif
>>
>>
>>
>>We could find an alternative way of constructing the
>>equivalent of an int 64 for the MAC address and its
>>formatting as a string for the UID.
>>
>>
>>
>>   Luis
>>
>>
>>
>>------------------------------------------
>>Bill Lorensen wrote:
>>
>>>Luis,
>>>I thought we would only update Cmake when absolutely necessary. Maybe now is the time.
>>>I have several nightlies that still use cmake 1.8. These are all linux or unix build I believe. I can update these to cmake 2.0. In the past, cmake upgrades on my GE linux and unix systems have wreaked havoc with a number of GE users, since they use my "production" cmake. The errors they encounter are usually due to holes that have been plugged in a cmake upgrade.
>>>If you think it is absolutely necessary, then I'll try to update to cmake2.0. Another problem is that I will be out of town for most of this week.
>>>Bill
>>>At 11:58 AM 1/23/2005, Luis Ibanez wrote:
>>>
>>>>Bill,
>>>>
>>>>Yeap, that seems to be the case.
>>>>
>>>>The complete story is that Mathieu is using a type "uint64" in
>>>>the encoding of the DICOM UIDs. This type is only available when
>>>>the compiler provides "long long" (which is a non-standard
>>>>extension), or actual 64bytes ints.
>>>>
>>>>In VTK a test is in place in order to let CMake find whether the
>>>>platform supports such types or not. I copied the tests from VTK
>>>>and set it up in ITK in order to conditionally compile that section
>>>>of code only when"long long" or true 64bytes ints are available.
>>>>
>>>>
>>>>However, the test requires some functionality available in
>>>>CMake post 1.8.
>>>>
>>>>
>>>>
>>>>Which raises up the question:
>>>>
>>>>
>>>>  What version of CMake are we requiring for ITK 1.10  ?
>>>>
>>>>
>>>>
>>>>I thought we were always requiring an ITK release to be
>>>>used with the latest release of CMake. (in this case 2.0.5).
>>>>
>>>>
>>>>Are we following that guideline for this release ?
>>>>
>>>>
>>>>(Maybe I'm missing a better way of dealing with the int64 type...)
>>>>
>>>>
>>>>    Luis
>>>>
>>>>
>>>>
>>>>----------------------
>>>>Bill Lorensen wrote:
>>>>
>>>>>Luis,
>>>>>Looks like there should be a:
>>>>>INCLUDE (${CMAKE_ROOT}/Modules/CheckTypeSize.cmake)
>>>>>but I think this may cause us to require a cmake after 1.8. I'm not sure.
>>>>>Bill
>>>>>At 09:01 AM 1/23/2005, Luis Ibanez wrote:
>>>>>
>>>>>>Bill,
>>>>>>
>>>>>>I'll look into the config problems on Terminus (Windows)
>>>>>>it seems that it has been running a CMake 1.9, that doesn't
>>>>>>support the recent check we added for int64.  I'll update
>>>>>>CMake in that machine.
>>>>>>
>>>>>>
>>>>>>
>>>>>>    Luis
>>>>>>
>>>>>>
>>>>>>
>>>>>>-----------------------
>>>>>>Bill Lorensen wrote:
>>>>>>
>>>>>>>Luis,
>>>>>>>I fixed gdcm problems on solaris.
>>>>>>>I fixed itkGDCMImageIO problem.
>>>>>>>Looks like config problems on Windows platforms. I'll have to leave that one to you.
>>>>>>>Bill
>>>>>>>
>>>>>>>_______________________________________________
>>>>>>>Insight-developers mailing list
>>>>>>>Insight-developers at itk.org
>>>>>>>http://www.itk.org/mailman/listinfo/insight-developers
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>_______________________________________________
>>>>>>Insight-developers mailing list
>>>>>>Insight-developers at itk.org
>>>>>>http://www.itk.org/mailman/listinfo/insight-developers
>>>>>
>>>>>
>>>>>_______________________________________________
>>>>>Insight-developers mailing list
>>>>>Insight-developers at itk.org
>>>>>http://www.itk.org/mailman/listinfo/insight-developers
>>>>
>>>>
>>>>
>>>>
>>>>_______________________________________________
>>>>Insight-developers mailing list
>>>>Insight-developers at itk.org
>>>>http://www.itk.org/mailman/listinfo/insight-developers
>>>
>>>_______________________________________________
>>>Insight-developers mailing list
>>>Insight-developers at itk.org
>>>http://www.itk.org/mailman/listinfo/insight-developers
>>
>>
>>
>>_______________________________________________
>>Insight-developers mailing list
>>Insight-developers at itk.org
>>http://www.itk.org/mailman/listinfo/insight-developers
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers at itk.org
>http://www.itk.org/mailman/listinfo/insight-developers