OK, I give up! I'm sure that there is a "right " way to do it that
will be "obvious, and easy to use", but I haven't found it.
I am able to do a respectable job of building across many platforms,
and getting the results posted to a Dart(2) dashboard.
But, I have two problems that I feel are not being addressed in an
adequate manner.
First, I would like to run a "pre-compile" in order to see just what
items are actually defined in a particular context.
With `gcc`, I am able to specify
-E
-dM
-c
-DCTEST_FULL_OUTPUT=1
and get some feedback about a particular compile environment.
However, I would like to be able to do this "in the actual context
that will apply to the build".
My other problem is that I have "tests" that are actually a number of
steps. In particular, to perform all of a test, I must:
Run a "pre-test" stage that copies "input files" to the current
directory and then runs a program to us those files to establish a
"known testing configuration" before I run the actual test.
Run the test, and collect its statistics, in particular CPU time. ( I
am NOT interested in the composite time, but rather, the time
associated with a particular configuration.
Evaluate the output in order to see if it meets expectations.
Failure to properly execute, of failure of the post-processing
segment, will result in a "failure" for the test.
However, and this is important because I am concerned with
"performance tuning", I really wish to report ONLY the time spent
during the primary "test", and ignore any additional pre- or post-
processing time.
Any suggestions?
Richard
_______________________________________________
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake
