replace parsing of $CLIBD/symop.lib with sgtbx

[rjgildea]

No description provided.

[graeme-winter]

Looked through the code changes, they all look sensible. Ran a test, seems to do what I would expect with the code in the branch. One concern I have is the handling of R3:H vs. H3 etc. - I know this can be a thing. Have you a test case for this?

[graeme-winter]

I do note that this now means fast_dp does not run on Python3 - possibly it could do? Maybe an absolute imports thing.

[graeme-winter]

Definite improvement, works for the use cases which were causing issues last week. Also tested with fast_rdp which seemed to do "the right thing"

[graeme-winter]

Travis test failure reflects previous lack of testing not a new problem - the issue is highlighted by the adding of tests.

[Anthchirp]

fast_dp will not run on python3 as it now depends on cctbx more than it did before.

[rjgildea]

I do note that this now means fast_dp does not run on Python3 - possibly it could do? Maybe an absolute imports thing.

fast_dp will not run on python3 as it now depends on cctbx more than it did before.

As far as I can tell fast_dp already depended on sgtbx in a number of places, e.g. when parsing the pointless xml output:

fast_dp/fast_dp/pointless_reader.py

Lines 27 to 33 in e84509d

	pointgroup = (
	sgtbx.space_group_type(lauegroup)
	.group()
	.build_derived_acentric_group()
	.type()
	.number()
	)

or when the --cell option is set:

fast_dp/fast_dp/cell_spacegroup.py

Lines 245 to 254 in fef9316

	def generate_primitive_cell(unit_cell_constants, space_group_name):
	"""For a given set of unit cell constants and space group, determine the
	corresponding primitive unit cell..."""
	uc = unit_cell(unit_cell_constants)
	sg = space_group(space_group_symbols(space_group_name).hall())
	cs = symmetry(unit_cell=uc, space_group=sg)
	csp = cs.change_basis(cs.change_of_basis_op_to_primitive_setting())
	return csp.unit_cell()

[rjgildea]

@graeme-winter r.e.

One concern I have is the handling of R3:H vs. H3 etc. - I know this can be a thing. Have you a test case for this?

$ dials.python
Python 2.7.16 (default, Apr  3 2019, 15:01:33) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-23)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from fast_dp import cell_spacegroup
>>> cell_spacegroup.check_spacegroup_name("H3")
'R 3 :H'
>>> cell_spacegroup.check_spacegroup_name("R3:H")
'R 3 :H'
>>> cell_spacegroup.ersatz_pointgroup("H3")
'R 3'

Is this the behaviour you expect?

[graeme-winter]

@rjgildea not sure, will review what it used to do and get back to you

[graeme-winter]

Reviewing

>>> cs.ersatz_pointgroup('H3')
'R3'
>>> cs.ersatz_pointgroup('R32')
'R32'
>>> cs.ersatz_pointgroup('H32')
'R321'
>>> cs.ersatz_pointgroup('H3')
'R3'
>>> reload(cs)
<module 'fast_dp.cell_spacegroup' from '/Users/graeme/svn/cctbx/modules/fast_dp/fast_dp/cell_spacegroup.py'>
>>> cs.ersatz_pointgroup('H3')
'R 3'
>>> cs.ersatz_pointgroup('R32')
'R 3 2'
>>> cs.ersatz_pointgroup('H32')
'R 3 2'

reload was checking out sgtbx branch - first 3 calls on master. Looks like the end game is more spaces and fewer rustic bugs, so happy.