the macros where used to make the influence of the valgrind calls zero
even for debug builds if valgrind client requests were
unavailable. Since this resulted in some inconsistencies, and the
performance hit is not terribly large, we now always use the same
inlined functions. For optimized builds the impact of those is still
zero...
This code is run unconditionally each time we do a reconfigure; if the
option is added at each time, they will accumulate (needlessly) on the
command-line, making it harder to inspect the log.
If the module root is not explicitly specified, the loading of its
configuration is delegated to a find module, which does a heuristic
search in known/surrounding directories. Most OPM modules use the
common logic in OpmPackage; this patch configures the search
specifically for opm-autodiff.
If we have checked out a newer Eigen3 directory in a sibling directory
to ours, assume that this is because the system version is obsolete and
that we want to use this one instead.
Eigen3 is a template-library, so we must compile the source code
directly together with ours instead of linking to a library.
If the build script gives us the location of a "built" Eigen3,
meaning a directory in which CMake has been run, locate the true
source directory from the cache entries.
If the package suite was given (e.g. DUNE_ROOT=/blum), then the code
set up the root for each individual package automatically (e.g.
DUNE_COMMON_ROOT=/blum/dune-common), but the path which was then
activated did not get the local build sub-directory (e.g. if we are
building opm-autodiff in /frub/opm-autodiff/build, then the local
build directory is "build/"), and thus this was not appended to the
library search path. The result was that the source was found (because
the root pointed to a valid source tree), but the library was not
(because it is "hidden" in the subdirectory).
because these classes should not have any side effects. case in point:
the warnings intended to be printed once get printed once per process
which can be quite a few times for parallel simulations with thousands
of cores. This could also be avoided by checking the MPI rank, but
IMHO this is way too much boilerplate code for a feature of questionable
value.
so far, it is only a flag which prevents the cache from being
deleted. To warn the user about a potential infinite loop, a warning
is printed if this flag was specified and some compiler or some compiler
flags were set by the user. This idea was proposed by Roland Kaufmann.
this seems to be a bug in cmake 2.8.10.2: if the user sets the
CMAKE_CXX_COMPILER variable for a build directory where this variable
has already been set, one gets an endless loop. This stings especially
if using the dunecontrol compatibility layer as the compiler flags are
unconditionally set via the CXX_FLAGS environment variable in the
options file. Running duncontrol on a module twice will thus trigger the
infinite loop if some compiler flags are set by the user.
The solution is relatively simple: Delete the CMakeFiles directory
before calling cmake. for the dunecontrol compatibility mode, we do
this in the configure script. For details about the cmake bug, see
http://www.cmake.org/Bug/view.php?id=14119
I haven't figured out why this happens, but on one of my computers
running kubuntu 13.10 g++ says the following if one tries to use
quadruple precision math:
/home/erne/src/ewoms/ewoms/common/quad.hh: In static member function ‘static quad std::numeric_limits<__float128>::min()’:
/home/erne/src/ewoms/ewoms/common/quad.hh:52:14: error: unable to find numeric literal operator ‘operator"" Q’
{ return FLT128_MIN; }
since this happens only on this machine and only with the c++ (but
not the C) compiler, i suppose that this is either an upgrade issue
with kubuntu or an issue with the ubuntu gcc stack. Strangely enough,
my other kubuntu machine does not have a problem with that...
In any case also testing for assignability of constants and using the
C++ compiler for the quadmath test won't hurt...
Some packages are split up in several library files, where each of them
doesn't have their own CMake find-module. So we cannot put this list in
the prerequisites. Unlike the headers, where we just add the directory,
all the libraries must be explicitly named on the link line. Thus, we
should allow to specify more than one name in the lib parameter.
The first name specified is designated as the "primary" library and is
assigned to the _LIBRARY variable; the others occur in _LIBRARIES as if
they were prerequisites.
Note that the build system cannot replicate this setup itself; it always
assumes that we are building (at most) one library from the source
files.
The resize() method for std::vector seems to copy an uninitialized
object for the new objects in the array on GCC 4.4. This means that we
should not assume that the copied objects are finalized in the
Parker-Lenhard parameters. Strangely enough this only seems to affect
older GCCs (at least 4.4) and only in debug mode...
this is intended to be used for ECLIPSE saturation curves, but the
code is more generic. (Be aware that using cubic splines for the
interpolation between sampling points is probably a better choice in
most cases.)
it seems to be implicitly included by newer compilers, but not GCC
4.4. In this case I agree with GCC 4.4...
Thanks to Bård Skaflestad for the hat-tip.
- add methods to calculate the derivatives of the capillary pressures
and relative permeabilities with regard to the phase saturations,
temperature, absolute pressure and phase composition
- extend the unit test to enforce the above
- make the NullMaterial conform to the API and add it to the unit test
- introduce Opm::NullMaterialTraits for material laws that don't use
any phase indices
They are now all generic capillary pressure laws, but with some
additional methods to make working with them easier. For this reason,
they have been moved up one directory to opm/material/fluidmatrixinteractions
Also, a unit test which ensures that all capillary pressure laws
conform to their respective APIs has been added.
