Adjust code to reflect comment.

Tobias Malkmus authored 9 years ago and Christoph Grüninger committed 9 years ago

This patch adds the cmake findpackage script which locates Tim Davis' SuiteSparsae collection.

If a certain setting of the SuiteSparse library should be used it can be found using
find_package( SuiteSparse COMPONENETS <list-of-components> )
with a specific list of components.
Existence of components can be check using the SUITESPARSE_<comp>_FOUND/HAVE_SUITESPARSE_<comp> cmake variable.

The UMFPack search routines are redirected to the more general FindSuiteSparse script.

We may drop the 'links' section in favour of just linking those pages
in the text.

Deprecate dune/common/alignment.hh

alignment.hh provides Dune::AlignmentOf<T>::value which is identical to std::alignment_of<T>::value.
This merge request replaces use of AlignmentOf with the shorter alignof() operator introduced in C++11 and deprecates the header.

See merge request !6

This is intended to contain everything related to type
computations, like, e.g., type traits, helpers for
overload resolution, ...

GitLab readable contribution guide, license and README

Adding the `.md` extension enables rendering of the files, which is much nicer.

Also ported the contribution guide from the User Wiki.

Renamed COPYING to LICENSE.md because that is the name that GitLab looks for to display it nicely on the project starting page.

If this one is accepted, I will do the same for the other core modules.

See merge request !5

The alignof() operator was introduced in C++11.

Conans ancient compiler has problems since the latest
changes to the Dune Exceptions. We switch the dune-common
check to a more easy header...

Implement some new features for dune_add_test

* `COMMANDS` was requested by @tkoch on the mailing list
* `COMPILE_ONLY` is useful for my latest YaspGrid tests, that check the code with and without deprecated interface features.
* `TIMEOUT` sets the timeout for ctest and is need in parallel testing

See merge request !1

This reverts commit 8bd9ff56.

The commit did not apply the correct fix, as MPI_RANKS is not a number
but a list of ranks instead. Also, I doubt the existence of ">".

ADDTEST_MPI_RANKS is set two lines above if it has been emtpy.

This allows to give priorities to otherwise ambiguous
overloads. It's main use is to make multiple overloads
work, when only some are ruled out via SFINAE. This is
e.g. used in the concepts check that I'll move to
dune-common soon.

The implemention is trivial but makes such applications
more readable and idiomatic compared to other approaches
where, e.g., null pointers are passed to overloads
have T* and void* dummy arguments.

Due to it's simplicity a separate header may not be justified,
but none of the existing seems to fit.

It needs to be specified when using MPI_RANKS. IT otherwise
defaults to 5 minutes. No test should ever run that long, I
guess, but the default keeps automated systems from entirely
hanging up.

It does not work the other way round, because gitlab is stupid.

Some whitespace changes included to make the more strict
whitespace hook happy (tabs removed).

Mostly C&P from the User wiki, with adjustments
to the new workflow.

Feature/dune exception derived from std exception

I almost missed Christian's last commit. It was poor luck  that I had a look into the commit history before replacing the old remote URL by the new, Gitlab-based one. I manually transferred his commit and saved myself the work of programming what Christian already did.

See merge request !2

Set if the given test should only be compiled during `make build_tests`,
but not run during `make test`. This is useful if you compile the same
executable twice, but with different compile flags, where you want to assue that
it compiles with both sets of flags, but you already know they will produce the
same result.

It allows to specify an arbitrary command to be executed
when running the test. This may for example be used to wrap
a script around a C++ executable.

Previously, the script that splits the in-module documentation
into parts exited with cryptic error messages that gave no hint
on where the typo happened. It now prints the offending line.

In dune-testtools we built a sphinx cmake documentation which is customized
quite a bit. With the core functionality being in dune-common, it is hard
to maintain it twice. I have therefore added some options to the dune-common
doc infrastructure, that are beneficial downstream.

You can now:
* provide custom rst sources
* provide a custom sphinx-build conf file
* limit the documentation to the current module, instead of adding all requirements automatically

The defaults of all those parameters result in an unchanged behaviour
(compared to before this commit).

As talked about on the developer meeting, a configure
switch should allow to build the tests during make all.
You can either set it through your opts file or activate
it on a per module level. If not active, you have to run

make build_tests && make test

to run the test suite.

We should always use the DUNE_ prefix for our cmake
variables, to avoid scoping clashes and to stay consistent.

Sorry for not doing this immediately, it was an oversight.