[vtk-developers] VTK without testing

[Bill Lorensen]

I see what you mean. I'll try a clean build from the command line.

If this works, I think there still is a bug. If I turn off testing after an
initial cmake, there should be no generated tests.



On Thu, May 23, 2013 at 3:13 PM, Bill Lorensen <bill.lorensen at gmail.com>wrote:

> Mine was a fresh build tree.
> ctest version 2.8.3
> cmake version 2.8.3
>
> I'll try with the latest cmake/ctest
>
>
>
> On Thu, May 23, 2013 at 2:53 PM, Marcus D. Hanwell <
> marcus.hanwell at kitware.com> wrote:
>
>> On Thu, May 23, 2013 at 1:53 PM, Marcus D. Hanwell
>> <marcus.hanwell at kitware.com> wrote:
>> > On Thu, May 23, 2013 at 1:11 PM, Bill Lorensen <bill.lorensen at gmail.com>
>> wrote:
>> >> Folks,
>> >>
>> >> There have been a few user comments about VTK6 regarding testing.
>> >>
>> >> I've started to look at the issues.
>> >>
>> >> My first attempt. Build VTK6 with BUILD_TESTING Off and BUILD_EXAMPLES
>> Off.
>> >>
>> >> VTK6 builds find. But I run ctest and VTK6 tries to run over 700
>> tests. Only
>> >> the header tests pass.
>> >>
>> >> Why is VTK6 generating any tests if BUILD_TESTING is Off. I suspect
>> some
>> >> automatic magic is the culprit.
>> >>
>> > I was going to take a look at this before you volunteered, I worked on
>> > some of this logic and would be happy to dig into it a little if Brad
>> > hasn't already figured it out ;-)
>> >
>> On further investigation I don't see the issues reported. The tests
>> you see are likely an artifact of CTest and CMake getting out of sync
>> Bill - if I do a default cmake run and then disable build testing then
>> ctest will try to run tests that don't exist. If I run CMake as
>> follows,
>>
>> cmake -G "Ninja" -DBUILD_TESTING:BOOL=OFF ~/src/VTK
>>
>> Then I see,
>>
>> $ ctest
>> *********************************
>> No test configuration file found!
>> *********************************
>>
>> As expected. The version of VTK I built I also linked to from
>> MongoChem, which uses the Qt and charting components in VTK with no
>> issue. I will dig into whether this is some break in logic when the
>> cache still has some testing settings hanging around from the initial
>> run of CMake. Looking at the CMake code it looks like all of the
>> testing calls are guarded correctly with if(BUILD_TESTING) ...
>>
>> Marcus
>>
>
>
>
> --
> Unpaid intern in BillsBasement at noware dot com
>



-- 
Unpaid intern in BillsBasement at noware dot com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20130523/805b563c/attachment.html>