[CMake] "How To" Question

[John Doe]

On Tue, May 13, 2008 at 3:27 PM, Richard Wackerbarth <richard at nfsnet.org> wrote:
> Juan,
>
>  Thanks for your suggestions.
>
>  Your "ADD_TEST" suggestion is just a (desirable) wrapping of the
> fundamental question which was
>
>  "Are there any standard CMake variables that can be tested to differentiate
> various machine capabilities?

Perhaps you can do something like this:

EXEC_PROGRAM(uname  ARGS -m OUTPUT_VARIABLE ARCH)

Where you use "uname" with the appropriate option or create your own
exec to come up with a version string.


>
>  As to the "Test A  && Test B", we need to see on the dashboard that "Test
> A" passed (with its accompanying metrics) and that "Test B" failed.
>  Any attempt to have the "combined test" reported as one entry will suppress
> valuable information.
>
>  Richard
>
>
>
>  On May 13, 2008, at 3:12 PM, John Doe wrote:
>
>
> > On Tue, May 13, 2008 at 3:00 PM, Richard Wackerbarth <richard at nfsnet.org>
> wrote:
> >
> > > I'm still trying to figure out how to handle the following two
> situations
> > > (simplified):
> > >
> > > When we run the tests for our Nightly Dashboard, in addition to a number
> of
> > > other tests, consider "Test A" and "Test B".
> > >
> > > If "Test A" fails, then I don't even want to attempt to run "Test B".
> > >
> >
> > You can make them part of the same test and then run them as:
> > testa.exe && testb.exe
> >
> > so that the second test won't run and they both show up under one test
> > name as failing.
> >
> > You can also wrap testb in a script so that it tests to see whether or
> > not the results of testa are available or even that your machine has
> > the appropriate resources..
> >
> > Perhaps you can script "ADD_TEST" in your CMakeLists if the test
> > doesn't apply to your platform.
> >
> > Regards,
> >
> > Juan
> >
> > >
> > >
> >
>
>
> >
> > > The program is a complex one that can run on a wide range of hardware.
> It
> > > can handle both "easy" data sets and "hard" ones. The "hard" ones push
> > > system resources, in terms of both cpu cycles and memory, and exercise
> parts
> > > of the program that are not necessary for the "easy" data sets.
> > >
> > > Although it would try, we know, in advance, that some of the older
> hardware
> > > is not "capable" in handling the more difficult data sets. The test will
> > > fail, either because of memory or time constraints.
> > >
> > > Therefore, I would like to avoid even attempting to run some of the
> tests
> > > on installations that have older CPUs or smaller memory.
> > > Are there any standard CMake variables that can be tested to
> differentiate
> > > various machine capabilities? Is there any interest in establishing
> some?
> > >
> >
>