relax SVN archive

Merged revisions 17017-17021 via svnmerge from
svn+ssh://bugman@.../svn/relax/trunk

........
r17017 | bugman | 2012-06-21 10:36:48 +0200 (Thu, 21 Jun 2012) | 3 lines

The spin ID lookup table has been made private so that it is not included in the save files.
........
r17018 | bugman | 2012-06-21 21:49:43 +0200 (Thu, 21 Jun 2012) | 5 lines

Fix for reading of BMRB CSA saveframe data - spin IDs generated from the data now have molecule info.

This is problematic only in certain edge cases, for example a multi-molecule entry.
........
r17019 | bugman | 2012-06-21 22:05:06 +0200 (Thu, 21 Jun 2012) | 6 lines

Removal of a hack from the generic_fns.bmrb.generate_sequence() function.

This hack was for naming unnamed spins. But this is not needed as the
generic_fns.mol_res_spin.create_spin() function already does this but with many more safety checks.
........
r17020 | bugman | 2012-06-21 22:18:14 +0200 (Thu, 21 Jun 2012) | 6 lines

Removed a hack from the generic_fns.relax_data.pack_data() function for the BMRB support.

This calls the generic_fns.bmrb.generate_sequence() function! As non-BMRB code paths access the
pack_data() function, this is a nasty hack which would have caused problems in the future.
........
r17021 | bugman | 2012-06-22 13:41:40 +0200 (Fri, 22 Jun 2012) | 26 lines

Creation of a fast molecule, residue and spin data lookup framework using private metadata.

This consists of two elements: The already existing private lookup table now at
cdp.mol._spin_id_lookup which is a dictionary with spin IDs as keys and a list of molecule, residue
and spin indices as values; and a set of private variables within the molecule, residue and spin
containers which identify the parent container names, numbers and indices. As all data is private,
it will not be visible to the user or be saved in the XML results and save files and should be
considered volatile.

All this private metadata is kept up to date via the two new generic_fns.mol_res_spin functions
metadata_prune() and metadata_update(). For fast operation, these methods can update specific
container subsets via the mol_index, res_index and spin_index arguments. All parts of relax which
modify the data pipe's molecule, residue and spin data structure (the generic_fns.mol_res_spin
functions and test suite) call these two functions as needed. Two auxiliary functions
spin_id_variants() and spin_id_variants_elim() have been added to create all possible matching spin
ID strings for a given spin (the second created IDs strings which should no longer exist).

The speed ups from this change are significant. On one system, the system and unit tests decrease
from 492.8s/26.4s to 434.3s/25.1s. On another the decrease is from 330.7s/17.4s to 258.9s/15.4s.

In addition, the pipe argument has been added to the generic_fns.mol_res_spin functions
create_molecule(), create_residue(), create_pseudo_spin() and create_spin(). Also, the molecule
name will now always be a string! Previously this was allowed to be an integer. This is needed for
the private metadata functions to operate correctly. A number of unit tests have been updated for
the changes.
........

Creation of a fast molecule, residue and spin data lookup framework using private metadata.

This consists of two elements: The already existing private lookup table now at
cdp.mol._spin_id_lookup which is a dictionary with spin IDs as keys and a list of molecule, residue
and spin indices as values; and a set of private variables within the molecule, residue and spin
containers which identify the parent container names, numbers and indices. As all data is private,
it will not be visible to the user or be saved in the XML results and save files and should be
considered volatile.

All this private metadata is kept up to date via the two new generic_fns.mol_res_spin functions
metadata_prune() and metadata_update(). For fast operation, these methods can update specific
container subsets via the mol_index, res_index and spin_index arguments. All parts of relax which
modify the data pipe's molecule, residue and spin data structure (the generic_fns.mol_res_spin
functions and test suite) call these two functions as needed. Two auxiliary functions
spin_id_variants() and spin_id_variants_elim() have been added to create all possible matching spin
ID strings for a given spin (the second created IDs strings which should no longer exist).

The speed ups from this change are significant. On one system, the system and unit tests decrease
from 492.8s/26.4s to 434.3s/25.1s. On another the decrease is from 330.7s/17.4s to 258.9s/15.4s.

In addition, the pipe argument has been added to the generic_fns.mol_res_spin functions
create_molecule(), create_residue(), create_pseudo_spin() and create_spin(). Also, the molecule
name will now always be a string! Previously this was allowed to be an integer. This is needed for
the private metadata functions to operate correctly. A number of unit tests have been updated for
the changes.

Removed a hack from the generic_fns.relax_data.pack_data() function for the BMRB support.

This calls the generic_fns.bmrb.generate_sequence() function! As non-BMRB code paths access the
pack_data() function, this is a nasty hack which would have caused problems in the future.

Removal of a hack from the generic_fns.bmrb.generate_sequence() function.

This hack was for naming unnamed spins. But this is not needed as the
generic_fns.mol_res_spin.create_spin() function already does this but with many more safety checks.

Fix for reading of BMRB CSA saveframe data - spin IDs generated from the data now have molecule info.

This is problematic only in certain edge cases, for example a multi-molecule entry.

The spin ID lookup table has been made private so that it is not included in the save files.

Many fixes for the model-free system tests and related data.

Now all model-free tests pass under the new interatomic data design.

The model-free data_init() method now sets boolean parameters to the default of False.

This excludes the selection flag which is set to True. The data_init() method no longer uses the
data_names() API method but the self.PARAMS.loop() method for returning the parameter names.

Improvements for the reading of old 1.2 relax results files for the attached proton spin containers.

Improvements and fixes for the generic_fns.relax_data.pack_data() function.

This affects all the relaxation data reading user functions.