[Insight-developers] In the spirit of the green dashboard
Bill Lorensen
bill.lorensen at gmail.com
Mon May 25 08:49:09 EDT 2009
Fortunately, we do not require CMake 2.6.
On Mon, May 25, 2009 at 8:40 AM, Luis Ibanez <luis.ibanez at kitware.com> wrote:
>
> Bill,
>
> Yeap, this seems to be the real source of the problem.
>
> My Cygwin build using CMake 2.4 completed linking the
> Numeric tests, which is where it was failing first due
> to that multiply defined symbols.
>
> We will report this to the CMake developers.
>
>
> The full Experimental build is still on its way...
>
>
> Luis
>
>
> --------------------
> Bill Lorensen wrote:
>>
>> My Experimental cygwin build: BillsLaptop win32-c++, is fine. There
>> are 2 failing tests, but I do have some local modifications.
>>
>> I'm pretty sure that the problem was caused by the cmake update.
>>
>> Bill
>>
>> On Mon, May 25, 2009 at 6:56 AM, Luis Ibanez <luis.ibanez at kitware.com>
>> wrote:
>>
>>> Bill,
>>>
>>> That looks like a thread worth-following.
>>>
>>> I'm using a CVS version (5 days ago) of CMake in order to build
>>> in darwinia.kitware.
>>>
>>>
>>> Let me start a build in darwinia using a CMake 2.4.
>>>
>>>
>>> Luis
>>>
>>>
>>> --------------------------
>>> Bill Lorensen wrote:
>>>
>>>> I am starting to suspect that the linker errors started after an
>>>> upgrade from cmake 2.4 to cmake 2.6.
>>>> The cmake icon for the last good cygwin build (May 7) says it used
>>>> cmake 2.4. The icon for the next day (May 8) says it used cmake 2.6
>>>> and May 8 is the day the linker errors began.
>>>>
>>>> Bill
>>>>
>>>> On Sun, May 24, 2009 at 11:27 PM, Bill Lorensen
>>>> <bill.lorensen at gmail.com>
>>>> wrote:
>>>>
>>>>
>>>>> I noticed that my cygwin version of cmake is 2.4. The dash14.kitware
>>>>> cmake is version 2.6. Can you determine when the cmake was updated on
>>>>> that system?
>>>>>
>>>>> Bill
>>>>>
>>>>> On Sun, May 24, 2009 at 11:20 PM, Bill Lorensen
>>>>> <bill.lorensen at gmail.com>
>>>>> wrote:
>>>>>
>>>>>
>>>>>> I'm wiping my binary tree. I'll try a fresh experimental tonight...
>>>>>>
>>>>>> On Sun, May 24, 2009 at 10:54 PM, Luis Ibanez
>>>>>> <luis.ibanez at kitware.com>
>>>>>> wrote:
>>>>>>
>>>>>>
>>>>>>> Hi Wes,
>>>>>>>
>>>>>>> All the builds that I have attempted so far with Cygwin + gcc344
>>>>>>> using
>>>>>>> Shared Libraries ON result in the same set of duplicate symbols.
>>>>>>>
>>>>>>> I attempted to repeat your experiment of going back in CVS time, and
>>>>>>> I
>>>>>>> manage to get back to January 1st 2009 and still get the same link
>>>>>>> errors.
>>>>>>>
>>>>>>> (Note that I have to update the two files to today
>>>>>>>
>>>>>>> /Insight/Utilities/gdcm
>>>>>>> CMake/FindUUID.cmake
>>>>>>> src/gdcmUtil.cxx
>>>>>>>
>>>>>>>
>>>>>>> to get over the uuid problem.
>>>>>>>
>>>>>>>
>>>>>>> I haven't find an explanation of why the build in dash14 is green on
>>>>>>>
>>>>>>> May 8th
>>>>>>> http://www.cdash.org/CDash/index.php?project=Insight&date=20090508
>>>>>>>
>>>>>>> and
>>>>>>>
>>>>>>> the link errors appear on May 9th
>>>>>>> http://www.cdash.org/CDash/index.php?project=Insight&date=20090509
>>>>>>>
>>>>>>>
>>>>>>> Despite the fact that locally, the link errors can be replicated as
>>>>>>> back as January 2009.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> It is interesting that in Linux, when building Shared, we produced
>>>>>>> a shared libraries
>>>>>>>
>>>>>>> libitkvnl.so
>>>>>>>
>>>>>>>
>>>>>>> while in Cygwin we produce a static library:
>>>>>>>
>>>>>>> libitkvnl.a
>>>>>>>
>>>>>>> and a Common library
>>>>>>>
>>>>>>> cygITKCommon-3.11.0.dll
>>>>>>>
>>>>>>> Records from the Developers mailing list indicate that for
>>>>>>>
>>>>>>> ITK 3.12: Dash14 was green except for a failing test
>>>>>>> ITK 3.10: Dash14 was green except for a failing test
>>>>>>>
>>>>>>>
>>>>>>> It is not clear if at that time it was using Shared ON.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Luis
>>>>>>>
>>>>>>>
>>>>>>> -----------------
>>>>>>> Wes Turner wrote:
>>>>>>>
>>>>>>>
>>>>>>>> Luis: I am not seeing a Darwinia build on the Experimental for today
>>>>>>>> or
>>>>>>>> yesterday. Did it pass?
>>>>>>>>
>>>>>>>> I am skeptical that the uuid fix will actually solve the problem we
>>>>>>>> are
>>>>>>>> seeing on Dash14 since it is entirely duplicate symbols with the vnl
>>>>>>>> libraries. However, given that Bill seems to have a successful
>>>>>>>> build,
>>>>>>>> I am
>>>>>>>> willing to believe it is a configuration problem on dash14 and not a
>>>>>>>> problem
>>>>>>>> with ITK. Would either if you be willing to submit a Nightly cygwin
>>>>>>>> build
>>>>>>>> with Shared=ON until I can sort out the issue? I would be more
>>>>>>>> comfortable
>>>>>>>> having both Shared=On and Shared=OFF represented anyway ...
>>>>>>>>
>>>>>>>> - Wes
>>>>>>>>
>>>>>>>> On Sat, May 23, 2009 at 9:26 AM, Luis Ibanez
>>>>>>>> <luis.ibanez at kitware.com
>>>>>>>> <mailto:luis.ibanez at kitware.com>> wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>> Mathieu,
>>>>>>>>
>>>>>>>> Thanks for the clarification.
>>>>>>>> That was very helpful.
>>>>>>>>
>>>>>>>>
>>>>>>>> Bill,
>>>>>>>>
>>>>>>>> Thanks for committing it.
>>>>>>>>
>>>>>>>> I just updated, and I'm now submitting another
>>>>>>>> Experimental build from darwinia.kitware.
>>>>>>>>
>>>>>>>>
>>>>>>>> Luis
>>>>>>>>
>>>>>>>>
>>>>>>>> ----------------------
>>>>>>>>
>>>>>>>> Bill Lorensen wrote:
>>>>>>>>
>>>>>>>> I just checked in a cygwin specific ifdef into gdcmUtil.cxx. I
>>>>>>>> have
>>>>>>>> had it on my local build for some time now and I guess I had
>>>>>>>> forgotten
>>>>>>>> to check it in.
>>>>>>>>
>>>>>>>> Bill
>>>>>>>>
>>>>>>>> On Sat, May 23, 2009 at 3:52 AM, Mathieu Malaterre
>>>>>>>> <mathieu.malaterre at gmail.com
>>>>>>>> <mailto:mathieu.malaterre at gmail.com>> wrote:
>>>>>>>>
>>>>>>>> On Sat, May 23, 2009 at 2:09 AM, Luis Ibanez
>>>>>>>> <luis.ibanez at kitware.com <mailto:luis.ibanez at kitware.com>>
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>> Wes,
>>>>>>>>
>>>>>>>> I just submitted the Cygwin experimental.
>>>>>>>>
>>>>>>>> It dies miserably due to the missing symbols
>>>>>>>> related to the UUID libraries.
>>>>>>>>
>>>>>>>> CMake finds:
>>>>>>>> UUID_LIBRARY:FILEPATH=/usr/lib/w32api/libuuid.a
>>>>>>>>
>>>>>>>> but it seems that this library in Cygwin doesn't
>>>>>>>> have all the symbols that GDCM is looking for.
>>>>>>>>
>>>>>>>> We may have to make some adjustments on the
>>>>>>>> CMakeList.txt file that search for the uuid
>>>>>>>> library.
>>>>>>>>
>>>>>>>>
>>>>>>>> UUID lib is *not* used on cygwin. If it isn this is a bug,
>>>>>>>> it should
>>>>>>>> simply use the UuidCreate symbol from rpcrt4.
>>>>>>>>
>>>>>>>> --
>>>>>>>> Mathieu
>>>>>>>> _______________________________________________
>>>>>>>> Powered by www.kitware.com <http://www.kitware.com>
>>>>>>>>
>>>>>>>> Visit other Kitware open-source projects at
>>>>>>>> http://www.kitware.com/opensource/opensource.html
>>>>>>>>
>>>>>>>> Please keep messages on-topic and check the ITK FAQ at:
>>>>>>>> http://www.itk.org/Wiki/ITK_FAQ
>>>>>>>>
>>>>>>>> Follow this link to subscribe/unsubscribe:
>>>>>>>> http://www.itk.org/mailman/listinfo/insight-developers
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Powered by www.kitware.com <http://www.kitware.com>
>>>>>>>>
>>>>>>>> Visit other Kitware open-source projects at
>>>>>>>> http://www.kitware.com/opensource/opensource.html
>>>>>>>>
>>>>>>>> Please keep messages on-topic and check the ITK FAQ at:
>>>>>>>> http://www.itk.org/Wiki/ITK_FAQ
>>>>>>>>
>>>>>>>> Follow this link to subscribe/unsubscribe:
>>>>>>>> http://www.itk.org/mailman/listinfo/insight-developers
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> Wesley D. Turner, Ph.D.
>>>>>>>> Kitware, Inc.
>>>>>>>> R&D Engineer
>>>>>>>> 28 Corporate Drive
>>>>>>>> Clifton Park, NY 12065-8662
>>>>>>>> Phone: 518-371-3971 x120
>>>>>>>
>>
>
More information about the Insight-developers
mailing list