The test started failing in commit 7d7f62e, but this was not detected
due to no automatic test environment. The commit changed the cell
numbering from "per-column (K,I,J)-ordering" to "per-plane
(I,J,K)-ordering". Consequently, the "correct_answer" seized to be
correct.
This change restores the "correct_answer" in the ordering introduced by
commit 7d7f62e.
While here, adjust style of the DisjointColumn test case for legibility.
Every program that relies on manual inspection has been moved to a new
(hopefully short-lived) directory called not-unit/; every remaining
file has been given the prefix test_ to indicate that this is the
executable test to be run.
If the program is built out-of-source-tree then the configuration file
(which is generated and not part of the source) will not be available
relative to the source code file.
Conflicts:
Makefile.am
opm/core/fluid/BlackoilPropertiesFromDeck.hpp
opm/core/fluid/SaturationPropsFromDeck.cpp
opm/core/fluid/SaturationPropsFromDeck.hpp
opm/core/fluid/blackoil/BlackoilPvtProperties.cpp
opm/core/fluid/blackoil/BlackoilPvtProperties.hpp
opm/core/fluid/blackoil/SinglePvtDead.cpp
This merge combines three more-or-less orthogonal features
for saturation tables: the option to use StoneII or Simple
three-phase behaviour, the option to fit a spline or not,
and finally setting the number of samples used (if spline
fitting).
Interfaces have changed, the most top-level one being
that BlackoilPropertiesFromDeck::init() now also takes
a ParameterGroup argument.
Introduced a simple fluid which has no problem with strange black oil behavior.
Intended for testing, but for now it is used in SaturationPropsFromDeck.
Recall that the class that used to be called SinglePvtDead has
been renamed to SinglePvtDeadSpline. If 'props_use_spline' is true,
that class is used (this is the default), which makes a monotone
spline that is uniformly, densely sampled. The new class simply
uses linear interpolation in the input tables.
This is a change from taking a vector containing the mapping to
deck-consistent logical cartesian indices. The mapping is contained
in the UnstructuredGrid::global_cell member, and may be null. The
change therefore saves the overhead of constructing a vector as a
copy of the data in the grid or (if null) as an identity mapping.
Fix a misunderstanding that's been present since the inception. The
LDFLAGS variable is entirely at the builder's disposal. Flags needed
for system-internal settings must go in the AM_* equivalents.
This is an experiment, not strictly sanctioned by the AX_LAPACK macro
documentation, to avoid linker errors related to missing symbol MAIN__
in the (GFortran) run-time support libraries. Strictly speaking, we
should be using the method outlined in the Autoconf manual, but we'll
try the easier way for now.
This commit is more or less equivalent to forward-porting commits
dd6e0fd3, c437eba2, and 9e885dc7 from the "ert" branch.
When OPM-Core is used as a Dune module (e.g., as enabled by commit
789bc5ca7), we stand the risk of creating multiple, conflicting
defintions of crucial <config.h> symbols (e.g. `HAVE_BOOST') and
build variables (e.g., `BOOST_CPPFLAGS' and `BOOST_LDFLAG').
Avoid this situation by prefixing these symbols with an `OPM_'
string such that Dune's `ENABLE_BOOST' solution works as intendend
in modules that are derived in the OPM context.
Use result of OPM_DYNLINK_BOOST_TEST to determine whether or not to
#define the Boost symbol BOOST_TEST_DYN_LINK .
While here, replace an unprintable character with the proper UTF-8
encoding in a comment.
Conflicts:
Makefile.am
opm/core/grid/cpgpreprocess/facetopology.c
opm/core/grid/cpgpreprocess/mxgrdecl.c
opm/core/grid/cpgpreprocess/preprocess.c
opm/core/grid/cpgpreprocess/sparsetable.c
opm/core/grid/cpgpreprocess/sparsetable.h
opm/core/grid/cpgpreprocess/uniquepoints.c
This merge brings a new, more resilient and feature-complete
corner-point processing. In particular, the new code features
exact, in-plane vertex coordinates for face nodes that arise as a
result of fault processing and which are not located on pillars.
Secondly, the resulting grid's cells are ordered lexicographically
with the I index cycling the most rapidly, followed by J and
finally K.
Finally, this merge also brings automatic handling of left-handed
coordinate systems which, until now, have produced negative cell
volumes as a result of face vertices being ordered such that
interface normals point from cell 2 to cell 1 in this case.
Left-handed coordinate systems are recognised using a simplistic
triple-product characterisation akin to the implementation of
function "processGRDECL" of MRST.
This code is now an (almost) exact replica of revision 1001 of
https://public.ict.sintef.no/viewvc/openrs/trunk/dune-cornerpoint/grid/preprocess/
The merge also removes opm/core/grid/cpgpreprocess/readvector*, so
remove tests/test_readvector.cpp (and accompanying Make rule) to
maintain a buildable tree.
Bård Skaflestad
Roland Kaufmann
Atgeirr Flø Rasmussen
Markus Blatt
Halvor Møll Nilsen
Joakim Hove