ALUGrid is needed to avoid unresolved symbols if dune-grid was
compiled with it. METIS is used by the parallel version of
ALUGrid. (Although it is not open source...)
The interface of the class PersistentContainer changed between DUNE
2.2 and DUNE 2.3 in an incompatible way, so we need to present
different versions of the class depending on the version of DUNE linked
with. This changeset adds the version variables for dune-grid to the
list of publicly necessary #defines that must go into using
dune-cornerpoint.
Notice that this will cause the definition of DUNE_GRID_VERSION_XXX
to be added to the flags list of dune-cornerpoint. This can be seen as
either a bug or a feature. If you try to link first dune-cornerpoint to
one version and then to another version of DUNE later, there is a very
good chance that you'll end up with problems anyway.
AGMG is now under a closed-source license, meaning that results
obtained with this solver is not freely reproducible by others.
Its use is therefore discouraged.
As of version 2.3, the DUNE AMG parts are competitive, so there
is a free and open alternative.
By returning a ${package}_SOURCES variable with the names of the code,
it will be propagated to ${project}_SOURCES and subsequently compiled
into the library.
AGMG (Algebraic Multigrid) uses this functionality.
Figure out where the closest dune.module is, and parse it for
information. The defines are added to config.h, so that our code
may behave differently for various versions available.
It actually works for opm-core 1.0 as well, but don't tell anyone;
we rather want people to use opm/core/version.h instead.
If a module has been backported and installed to a directory which
also contains the version number, e.g. /usr/include/dune-2.3, to
avoid clashing with an existing version in /usr/include, then this
part is now stripped from the include directory and added back to
the lib directory where we are looking for dune.module
This hook together with the bundled convenience macro makes it
possible to add the version of another module to config.h using
code like this:
macro (config_hook)
opm_need_version_of ("dune-common")
endmacro (config_hook)
The config variables in _CONFIG_IMPL_VARS are only written to config.h
and not the .pc or -config.cmake files. They can thus only be used by
the implementation files and not the headers. Use this set to avoid
exposing the variables to other projects, when it is likely that a
probe used by that other project may turn up a different result.
If you run `cmake --build . -- distclean` all generated files will
now be gone! This adds the files that have been introduced in later
improvements to the build system.
Clang aims to be compatible with GCC when it comes to command-line
parameters. Where we enable functionality based on the presence of
a GCC-compiler, we can use the same functionality with Clang.
This patch introduces a new variable CXX_COMPAT_GCC which is true if
the compiler handles the same options as GCC, and this variable is
subsequently used in tests instead of CMAKE_COMPILER_IS_GNUCXX (which
remains if we need to test if we really have GCC, e.g. for version)
The prerequisites are now declared in their own file, so it can be
included by the main system. There is thus only one canonical place
where the list of prerequisites exist.
Other modules (notably the precompiled header one) may set properties
directly on the source file; we need to separate them into categories
for which language they belong to.
The version declared for the build system (in dune.module) is checked
against what is in the source (e.g. opm/core/version.h) and if these
don't match, then issue an author warning.
This will help us keep the two version numbers in sync., since it
becomes very obvious when we don't. The benefit for this is to not have
the build system start mucking with the code just to dump some static
information in there.
When checking out from source code, the paths that were used to
configure the project should be carried on to the installed binary,
so the same libraries are used without much effort.
By using RUNPATH instead of (just) RPATH, it can be overridden with
LD_LIBRARY_PATH if the user so desires.
If we write a blank value into the cache, we tell CMake that ERT isn't
found and it won't check again. If we delete it from the cache in case
we didn't find the directories, then CMake will run the find module
again the next time we configure. Just probing for directories is cheap
so this is no problem. This allows us to install ERT, do a reconfigure
and it will then be picked up by the build system.
If we find ERT and it doesn't compile properly, a blank value will still
be written, and it won't try again the next time, so you must wipe the
cache if you *upgrade* ERT into a newer and more compatible version.
It gets tiresome to rebuild the project everytime one makes changes to
any of the unit tests. We don't likely publish results based on the
debug code anyway, so tracking the version number there is not such a
pressing issue.
Otherwise, the library may be declared "done" before the version number
process have had any change to touch files it depends upon. This ensures
that the version number is correct before we start building the library.
Akin to config.h, we write project-version.h whenever the VCS sha hash
of the project changes. This file can then be included to embed this
into the project. Since this changes more frequently, we choose another
file than config.h. Care is also taken to not rewrite the header if the
information doesn't change, but the project is rebuilt.
We cannot generate dune.module because that file is read by dunecontrol
before the build starts. It therefore make sense to let it be the
original and let the build system read it instead of duplicating the
This module is a adaption of Andreas Lauser's OpmParseDuneModule
in opm-material, modified to handle the more sparse info provided by
the other modules as well.
It is only used in one place, where it can easily be replaced by its
expansion. The reason for removing it is that it depends on the name
of the project, and I intend to read this using a routine it the very
directory it is including, hence the need for a different approach.
The linker that is used (ld vs. gold for instance) is often hidden
when using the compiler as a front-end (to include the correct
runtime libraries).
These functions enables us to probe the linker version and reports
its finding in the log.
Developers that are working to familiarize themselves with OPM will
probably start out in CMakeLists.txt to figure out where the project
declare its dependencies, thus the list should be referred to from
there instead of inside one of the sub-modules.
The hooks are meant to be project specific, and should therefore
either know their own name or at least have access to a variable with
that name in it.
This file can be included both in the project itself, and also in the
find module for the library, so that the list of dependencies can be
maintained in only one place.
This let you specify a bunch of directory names on the command line
to *all* packages, without getting annoying warnings that they are
defined but not used.
When using LTO, the linker calls back to the compiler to figure out
where a certain symbol is defined. However, in many versions there
is apparently a bug that is triggered if a template in inlined in
both a used library and in the main program, but with different
versions of the compiler. The Boost exception class is particular
prone for this.
Thus, we disable the -lto from the development builds by default;
if anyone wants to test of the behaviour of their own compiler, they
must now explicitly turn on this flag.
By setting EIGEN3_ROOT, we can point directly to the path of the
library checkout we want to use. No other searching will then take
place.
Notice that if you have found an Eigen3 installation in a previous
configure, it will not find a new version even if you introduce the
EIGEN3_ROOT variable!
