REGRESSION TEST

If your system is configured and you compiled everything
just try the tests with
> make

To do many tests in parallel, as usual
> make -j 4

HOW TO ADD A NEW TEST

Add a new directory rtXXX, let's say rt105, and go there:
> cd rt105

Put a standard makefile there:
> echo "include ../scripts/test.make" > rt105/Makefile

Put in the rt105 directory a "config" files containing the
options which define your test.
Presently, you can try one of the following:

(A) testing with simplemd
type="simplemd"

(B) testing with the driver. the arg variable allows to pass command
    line arguments to the driver
type="driver"
arg="--plumed=plumed.dat --timestep=0.05 trajectory.xyz"

Put in the rt105 directory all the input files that you need.
Typically:
  for simplemd: in input.xyz plumed.dat
  for driver: trajectory.xyz plumed.dat
Many tests share can share the same trajectory file. Just add
to the config file a line similar to
extra_files="../trajectories/name-of-the-file"
to include a common trajectory file

Make:
> make

Copy from the newly generated tmp directory the files that you
want to check and rename them into xxx.reference
> cp rt5/COLVAR COLVAR.reference

Add to the git repository all the files in the rt105 directory,
but for report.txt and tmp directory:
> git add .
(this should work since report.txt and tmp are git-ignored)
In this manner, other people will be able to perform the test

HOW TO RESET A TEST
If you get an error from a test but you think that the new
result is the correct one, just go to its directory and type
> make reset
Then commit the modified references


