New scons build option introduced to replace 'system_googletest'
and allow to not use 'googletest' module while running test
(all tests that require this module could be omited)
New option description:
'googletest',
"""Select whether to use gtest/gmock from system
installation ('system'), from a Git submodule ('submodule'), to decide
automatically ('default') or don't look for gtest/gmock ('none')
and don't run tests that depend on gtest/gmock.
If this option is set then it suppress the deprecated 'system_googletest' option."""
Old option is still presented and mentioned as 'deprecated'.
Option 'googletest' supresses the old 'system_googletest' one.
Update and make more consistent the specification of Python package
building. Since SCons can be run by Python 3 now, we cannot assume that
the Python running SCons is Python 2. This changes a bunch of
assumptions in SConstruct about where things should be built or
installed. This commit addresses those assumptions by making the options
for Python 2 and Python 3 symmetric.
If no full or minimal Python interface is being built, copy the minimal
interface into the build directory and use the sys.executable to run it,
so the tests that require CTML or CTI conversion can run.
The 'install' and 'test' targets had some undeclared dependencies on the 'build'
target, such that running 'scons install' or 'scons test' without having first
run 'scons build' would result in incomplete installation or test failures,
respectively.
Fixes#432.
The old version of this test didn't actually check any meaninful results. The
ChemEquil and VCS solvers are both able to solve this problem, but the
MultiPhase solver currently fails. Also, the BasisOptimize function may fail
depending on the order that the elements are specified in the input file.
On Linux systems, link tests against the Cantera shared library, rather than the
static library to reduce the size of the compiled test suite. This saves over 1
GB of disk space when compiling with debug symbols enabled, and also reduces the
run time of the test suite.
This does not work on Windows, where the shared library only exports the Cantera
C interface.
Use the correct platform-independent path separator instead of assuming that it
is ':'.
Determine the PYTHONPATH variable in a single place for each Python version.
The user's environmental variables are now passed down to the tests. Previously they had been ignored.
All of the python tests (minimal and full) had been failing for over a year or so for any installations which actually use the
LD_LIBRARY_PATH and PYTHONPATH environmental variables, which pretty much consists of all multiuser linux
machines.
If a gtest program crashes, register that as a single test failure.
Previously, the entire test suite would be aborted because the
attempt to open the test results file (gtest-foo.xml) would fail.
When instantiating a phase from a .cti file, do the conversion in memory,
without writing the XML representation to disk. This eliminates the unrequrested
XML files that Cantera normally generates, and also avoids errors when running
Cantera from a directory where the user does not have write permissons.
Took out single_species compilation option
took out some deprecated warnings
Fixed an error in copy constructors for thermo.
Still an error with PYTHONPATH and the test suite
-- more to come.
This allows use of more extensive warning settings for the code that's actually
part of Cantera without generating excessive warnings on code automatically
generated code (e.g. f2c) or code that isn't part of Cantera proper (e.g. gtest,
libexecstream).
