Hi,
For something small like this, it shouldn't be a problem. It's only
touching a function which is specific to your XYZ file reading code,
affects no other parts of relax, and is only in the test suite. So in
this situation, it should be ok. If you are interested in making
bigger changes, even in the future, then the relax developers can have
a private vote to accept you as a relax developer with full access
privileges. Apart from your mistake with svnmerge, your coding skills
are quite high. What do you think?
Regards,
Edward
On 11 August 2011 15:06, Han Sun <hasu@xxxxxxxxxxxxxxxxx> wrote:
Hi Edward,
You mean I should change the 1.3 line? Do I have the authority to do it?
Best
Han
On Aug 11, 2011, at 3:01 PM, Edward d'Auvergne wrote:
Hi,
That sounds good. Would you be able to make the change?
Cheers,
Edward
On 11 August 2011 14:42, Han Sun <hasu@xxxxxxxxxxxxxxxxx> wrote:
Hi Edward,
I checked out the copy and I think the problem can be very easily fixed.
The reason for this failure is because the system test code of
'test_read_xyz_internal2' is old. In this old test it assumes that the
vector is not a unit vector, which is however not identical with the
default
setting in the 'extract_vector' function. So to solve this problem you
just
need to replace the following code for extracting a vector in the
'test_read_xyz_internal2':
# Extract a vector between first two spins.
self.interpreter.structure.vectors(attached='@10', spin_id='@2')
self.assertAlmostEqual(a.bond_vect[0], -0.4102707)
self.assertAlmostEqual(a.bond_vect[1], 0.62128879)
self.assertAlmostEqual(a.bond_vect[2], -0.6675913)
Best wishes,
Han
On Aug 11, 2011, at 11:30 AM, Edward d'Auvergne wrote:
Hi Han,
I have now merged your xyz branch into the main 1.3 line. There was
only one conflict in this process, and it was easy enough to fix. I
have now run the test suite, and noticed that the
test_read_xyz_internal2 system test fails. This can be quickly
checked with:
$ ./relax -s Structure.test_read_xyz_internal2
I think I remember seeing this problem before, but I thought that it
was solved in the xyz branch. Could you check the 1.3 line and see if
all your code is in there? Or check what could be the source of this
problem? I would like to release a new version of relax soon with
this code (along with a series of bug fixes) prior to releasing a new
version with Gary Thompson's multiprocessor code, and then a version
with the new GUI code (hopefully all by the end of next week).
Cheers,
Edward