This is the Python super-compatibility release - relax will now run on Python 2.3 all the way to Python 3.3, a span of over 9 years of Python development. Some exceptions are Python 3.0 as numpy and scipy are not compatible with that release, the execution of external programs in Python 2.3 not being supported, and the GUI currently not available on Python 3 due to the wxPython project not being compatible with this new Python design yet. A number of important bug fixes are also included. For more details please see below. The new relax versions can be downloaded from http://www.nmr-relax.com/download.html. If binary distributions are not yet available for your platform and you manage to compile the binary modules, please consider contributing these to the relax project (described in section 3.6 of the relax manual, http://www.nmr-relax.com/manual/relax_distribution_archives.html). The full list of changes is: Features: * Python 2.3 to 3.3 support, including large changes throughout the code to be able to run on Python 3. * Improvements to the GUI tutorial in the model-free chapter of the relax manual. * A number of bugfixes. * Small improvements overall. Changes: * The scons 'clean' target now removes the Python 3 __pycache__ directories. * Small edit to the installation chapter of the user manual. * Decreased the Python version dependency from 2.5 to 2.3 in the installation chapter of the user manual. * More error checking for the associate_auto() method of the data pipe editor window. * Added data pipe bundle error checking for the GUI pipe editor window associate_auto() method. * Added some error checking for the data pipe bundle in the auto model-free analysis GUI code. * Added some special RelaxErrors for data pipe bundles. * Added some bug catching code for the observer objects. In some rare cases a registered method's key was set to None. This is now caught and a RelaxError thrown to prevent later indecipherable errors. * The setup.py application building script now complains if the Python setuptools are not installed. * Updated the relax prompt mode figure in the intro chapter of the user manual to the more modern prompt. * Improvements to the API documentation compilation. The excluded files and directories, as well as hidden ones, are no longer included in the list of files/directories to add to the documentation. * Added a file with the relax user functions used for the prompt screenshots. This is for the manual and the website. * Added the public domain LaTeX nth.sty style file for the user manual. Some LaTeX distributions do not have this style file and, as it is public domain, it can be legally distributed with relax allowing the PDF manual to compile on more systems. * Fixes for weird print statements with double brackets generated by the 2to3 Python conversion script. * Removed a debugging printout. * Python 3 fixes for one of the test data scripts - print statement with function call replacements. * Python 3 fixes for non-used Python code - converted print statements to function calls. * Python 3 fixes for the script for generating plots of magnetic field lines. * Another print statement to function Python 3 fix for the user manual. * The Python print statements in the user manual are now function calls to be Python 3 compatible. * Python 3 fix for the generic_fns.structure.geometric.angles_regular() function. Integer divisions no longer produce integers. * Better formatting of the test suite summary. * The text relax controller log is no longer cleared when a reset occurs. This allows the test suite results to still be presented in GUI mode. * Even cleaner exiting of the GUI - the interpreter thread is terminated by the exit_gui() method. * The GUI is now cleanly exited with a call to wx.App.ExitMainLoop rather than wx.Exit. * Python 3 fix for the compat module - the Queue2 object needs to always be defined. * Added support for Python 2.2 and earlier for the compilation of the C modules. * Removed an unused import of the Queue module from the multi-processor. * Python 3 fix for the ScientificPython PDB reader unit tests. The order of the keys returned by a dictionary's keys() method changes randomly in Python 3, so now they are sorted prior to comparison. * Redesigned the reset user function backend. This now no only clears out the relax data store, but it also resets the GUI if present. Some of the reset code comes from the tearDown() method of the GUI tests. All windows but the main GUI window are closed and the relax controller gauges are set to zero and the log window text cleared. These changes should allow GUI tests after an error or failure to pass, something which is currently problematic. * Disabled the initial relax intro printout from the GUI when running the test suite. This prevents the intro text from appearing in the first failed test. * Fix for the Mf.test_read_results_1_3_v2_broken() system test for Python 3.2. The object comparison method no longer converts dictionaries to strings for the comparison, as the string version is different in different Python versions. * Fix for the Mf.test_write_results() system test for the Python 3 versions. The logic for determining Python 3 versions was broken and the incorrect files was used for Python 3.1. * Better Python 2.3 support. The compat module is now imported at the very start to allow the builtins to be set before any other imports. The sorted() builtin method is now mimicked and the os.devnull string set for Python 2.3 and earlier. * Fix for the Mf.test_write_results() system test for Python 3.1. The XML version in Python 3.1 is the old style. Therefore the old results file is being used to check this Python 3.1 result. * Small improvements to the multiple Python version test suite testing script. * Reactivated support for Python 2.3. This mainly skips the missing 'subprocess' module. This however decreases relax's functionality a little. * Created a special script for testing out relax with Python versions 1.0 all the way to 3.3. This builds the C modules for each Python version in ~/bin and then runs the test suite, outputting everything to log files. * The Results system tests are no longer dependent on the relaxation curve-fitting C modules. This allows these tests to run when the module cannot be imported. * Python 2.5 and lower fix for the test_write_protein_sequence() unit test. The byte array is wrapped in an eval() statement to allow Python 2.5 and lower to parse the code without failing, and the byte array comparison is now only used for Python 3+. * All system and GUI tests reliant on the relax-fit C modules are deactivated if import fails. This removes a pile of useless error messages from the test suite and presents a table of skipped tests at the end. * More Python 3 fixes for the use of now non-existent string module functions. * Python 3 fix for the model-free BMRB export - many string module methods no longer exist. * Mass conversion of the alignment tensor data structures to the same new design as the diffusion tensor. This large set of changes matches all of those revisions for the diffusion tensor already committed. The alignment tensor data structures are now read only, and can only be modified via the set() method. This is a much simplified design which works on all Python versions. * Small clean ups of the diffusion tensor data structure code. * Deleted the now unused _update_sim_set() method of the diffusion tensor data structure. * Removed the now unused _update_sim_append() method from the diffusion tensor data structure. * Cleaned up the docstring of the diffusion tensor data object __setattr__() method. * Updated all of the diffusion tensor unit tests to the new design. * Fix for the reading of model-free results files from relax 1.2 when simulation data is missing. * Fix for the reading of relax 1.2 model-free results files for the diffusion tensor structure redesign. * Another fix for the fold_angles() diffusion tensor function - again an incomplete design conversion. * Fix for the setting of the diffusion tensor parameter errors in the model-free specific analysis code. * Fix for the setup of the model-free Monte Carlo simulations for the new diffusion tensor design. * Another fix for the diffusion_tensor.init user function - it was not completely converted. * Fix for the fixing of parameters in the model-free analyses. The diffusion tensor set_fixed() method is now used. * Fix for the XML output of the diffusion tensor - only the modifiable parameters are output. This was the previous behaviour and is needed for the test suite to pass. * Converted the palmer.extract user function to use the new diffusion tensor design. * The diffusion tensor bmrb_read() function now uses the set_fixed() method instead of fixed(). * The fix user function now uses the diffusion tensor set_fixed() method. * Renamed the diffusion tensor fixed() method to set_fixed() to avoid clashing with the 'fixed' object. * Fix for the model-free specific analysis duplicate_data() method for the new design. The diffusion tensor __mod_attr__ object is now called _mod_attr. * Fix for the diffusion tensor to_xml() method for the new design. For some reason the methods of the Element class are no longer blacklisted. * Converted the diffusion tensor data structure from_xml() method to the new tensor design. * Fix for the Diffusion_tensor.test_copy system test - the simulation parameters are now read-only. Instead, the diffusion tensor set() method needs to be called. * The setting of list values for the DiffTensorSimList object now works correctly. The private _set() method now works correctly by calling the base class method, and the normal setting of diffusion tensor simulation values produces a RelaxError. * Fix for the diffusion tensor __deepcopy__() replacement method for the new design. * The model-free specific analysis _disassemble_param_vector() method now uses the new diffusion tensor design. * Modified the setUp() method for the diffusion tensor system tests to use the new design. * Redesigned how diffusion tensor simulation structures are handled. The design is now much cleaner and works with all Python versions. * Removed all the unused imports from specific_fns.model_free.main. * A number of private diffusion tensor objects and methods have switched to the single leading '_' format. * Improvements to the diffusion tensor set() method. The parameters, errors and simulations are now properly differentiated and stored. * Converted the old diffusion tensor __setattr__() method into the set() method. This is the only way in which diffusion tensor parameters, errors and simulations can be set. * Renamed the diffusion tensor data structure type() method to set_type(). This is because the type is stored as the 'type' object, clashing with the method name. * Created the diffusion tensor data structure type() method for setting the tensor type. This is to remove the "cdp.diff_tensor.type = 'x'" code from the core of relax, as the structure is now read only. * The new diffusion tensor fixed() method has been created to allow the fixed flag to be changed. * Fix for the initialisation of the diffusion tensor data structure, now that it is read-only. * The diffusion tensor data structure has been completely converted into a read-only structure. The __setattr__() method now will always raise a RelaxError, and the diffusion tensor simulation data structure objects __setitem__() method will raise the same error. * Updated the relax version numbers and 'trunk' used relax user manual. For example the information about checking out the main development line was still talking about 1.3 rather than the trunk. * Python 3 fix for the setting of diffusion and alignment tensor simulation values. The previous code somehow worked in Python 2 but was not formally correct and broke in Python 3. * Python 3 fix for the model-free results file reading tests. The ordering of dictionaries is different in Python 3, so now these are properly converted from strings to dictionaries before comparison. This was not happening because of the XML changes from Python 2.7.3 onwards. * The relaxation curve-fitting system tests are now skipped if the module is missing or broken. This improves the printouts from the test suite and shows a summary of skipped tests rather than a pile of traceback messages and errors. * The message about skipping the GUI tests due to wxPython being missing is now more specific. This was being shown for all runs of the test suite when it only needs to appear if GUI tests have been run. * Added a Python 3 version of the truncated OMP model-free results file. This was created with trunk. * Removed the Python 3 byte array hack which should have been removed earlier. * The OMP model-free results file generation script now outputs for any relax version. * Python 3 fix for the Mf.test_latex_table system test. The latex_mf_table.py model-free system test script docstring contains backslashes, so the raw string format r"""Text""" is now used. * Python 3 support for Modelfree4 and Dasha. The subprocess.Popen class works with byte arrays rather than strings in Python 3+. The Python objects are now interconverted when the Python 3 encode() and decode() methods are detected. * Removed the pickle format information and arguments from the state user function definitions. * Eliminated the State.test_state_pickle() system test as pickled states are no longer supported. * Removed the ability to save and restore states using the pickle module. A pickled state is of no use to relax anymore. It's removal is needed for Python 3 support. So now everything defaults to the XML formatted output. * Python 3 fix - removed the use of the string module from generic_fns.spectrum. * Python 3 fix for the relax_io.open_write_file() function. This now matches the behaviour of open_read_file() in that there are three different behaviours for opening bz2 and gz files for writing to for the different Python versions (one for Python 2, one for Python 3.0 to 3.2, and one for Python 3.3+). All byte streams have been eliminated as open_write_file() is for creating text files. * Python 3 fix for the Noe.test_noe_analysis() system test for the grace.write precision changes. * For consistency between Python 2 and 3, the grace.write user function outputs to 15 decimal places. This increased precision will only be of use in the relax test suite. * Python 3 fix for the Pipes.test_pipe_bundle() system test. The order of bundle names returned by generic_fns.pipes.bundle_names() is not guaranteed in Python 3. * The C module compilation testing script now accepts the Python version as a first argument. * The relax_io.open_read_file() now supports all Python versions over 2.4. This required some really nasty hacks for Python 3.0, 3.1 and 3.2 with the Bzip2Fixed and GzipFixed classes overriding the incomplete and buggy bz2.BZ2File and gzip.GzipFile modules, and being wrapped around io.TextIOWrapper(). * Added the IO module to the relax information printout and dependency checks. * The manual C module compilation script is now executable. * Renamed the 'scripts' directory to 'devel_scripts' so that users are less likely to ask about the scripts. * Finished off the C module compilation testing script. * Added a script for testing out the C module compilation on multiple Python targets. * The relax_fit specific analysis module now supports both Python 2 and 3. * The relaxation curve-fitting C module now supports compilation on both Python 2 and 3. * Created the simple Sequence.test_sequence_copy() system test to catch bug #20213 (https://gna.org/bugs/?20213). * The Mf.test_bug_20213_asn_sidechain() system test now uses a temporary directory for output. * Added the Mf.test_bug_20213_asn_sidechain() system test to catch bug #20213. The data and script comes from the files 'sh3-47.2.zip' and 'run.py' attached to the bug report https://gna.org/bugs/?20213. The PDB now only contains Asp47, the optimisation parameters have been made almost insignificant, and all models but 'tm0' have been removed from the analysis. * The Python 3 dictionary values() method no longer returns a list, so a list() call is needed. * Python 3 bug fix for the geometric structure module - another integer division to float problem. * The Mf.test_write_results system test can now select the correct file to compare against in Python 3. The algorithm for determining if the 'final_results_trunc_1.3_v2' or 'final_results_trunc_1.3_pre_py2.7.3_v2' file should be used could not handle Python 3. * Python 3 fix for the format detection of results and save files. * Python 3 import fixes for the generic_fns.structure package using relative paths. * Python 3 fix - removed the use of the string.lower() function in the OpenDX mapping code. * Python 3 fix for the frame order system tests. As float to string conversions behave differently, the %.1f formatting is used to force only a single decimal place float. * Python 3 fix for the frame order system tests - float to string conversions behave differently. Now the explicit %.1f formatting is used to force only a single decimal place float. * Python 3 integer division to float fix for the frame order analysis. * Python 3 bug fix for the frame order analysis - another int division problem. * Python 3 fixes - eliminated all usage of the dictionary has_key() calls as they are no longer present. * Python 2 and 3 support in the generic_fns.relax_data module using 2to3. One print call was fixed after running 2to3. * Python 3 bug fix for the Structure.test_read_pdb_mol_2_model_scientific system tests. This is again an integer division problem returning a float. * Python 3 fix for the test_write_protein_sequence() unit test. This is again a string verses byte verses unicode problem. * Python 3 fix for the user function docstring creation in the prompt UI mode. Again this is the problem of a division now returning a float rather than an int. * Python 3 bug fix for the N-state model target function setup. The num_tensors variable needs to be an integer, but the Python 3 division will create a float type. * Python 3 fix for the results.read user function matching that of state.load. * Python 3 bug fix for the relax_io.read_spin_data() function. The built in max() function cannot handle the value of None, therefore the filter() function is used to remove all instances of None from the list. * Python 3 bug fix for the state.load user function. The header line of pickled states (rather than the standard XML states) is of the b'' byte format. This is now converted to a string, and the search expression is comparing it to the raw string r"<\?xml". * Better support for both Python 2 and 3 in the relax data store. The 2to3 script was used on all of the files in the data package. * Python 3 preparation - the relax data store (the data package) now supports both Python 2 and 3. * Python 3 fix - the relax_errors.AllRelaxErrors object is now a proper tuple. Due to bad coding, it was previously a nested tuple. This nested tuple worked in Python 2, but is fatal for Python 3. * Python 3 fixes - the character '\' is now properly escaped as '\\' in the stereochemistry auto-analysis. * Fix for the test suite summary for Python 3. The test suite now runs, but fails miserably, under Python 3. * Fix for the running of the test suite under Python 3. The zip() function used in the loadTestsFromTestCase() function is now an iterator, so it needs to be passed through the list() function to generate a list. * Fix for the test_parse_token_multi_element_name() unit test, as parse_token() no longer sorts. * Python 3 fix for the generic_fns.mol_res_spin.parse_token() function. Mixed lists for int and string can no longer be sorted. This sort call is not needed anyway. * Automatically converted the generic_fns.mol_res_spin module to support both Python 2 and 3. * For running relax with Python 2, the __builtin__.range() function has been replaced with xrange. This causes large speed ups (speed that was lost with the earlier xrange() to range() conversions), and memory decreases. For example on one system, the system test time decreased from 513.029s to 487.586s. * The compat module now has the py_version variable specifying if this is Python 2 or 3. * Import fix for the OpenDX mapping package, recently broken with the relative import for Python 3 change. * More usage of the is_unicode() function in the generic_fns.mol_res_spin module. * Created the check_types.is_unicode() function for Python 2+3 compatibility. This is used in the generic_fns.mol_res_spin module. * Another raise() function call to statement change for 2to3 preparations. * Converted some raise() function calls to raise statements in preparation for the 2to3 conversion. * Converted the ScientificPython PDB reader to support both Python 2 and 3. The __repr__() method was manually modified due to the 'list' variable clashing with the 'list'type. * Created a new module for simultaneous Python 2 and 3 support in relax called 'compat'. * Python 3 fixes - the list() function is now used in combination with range() to generate the ordered list. range() in Python 3 is an iterator object (just as xrange was), so now the list() function has to be called. * Python 3 preparations - mass conversion of all xrange() calls to range(). * Created the special check_types.is_filetype() function for checking for files in all Python versions. * Python 3 - eliminated an unneeded xrange call. * Python 3 - eliminated the use of the map() function, as this behaves differently in Python 3. * Python 3 - removed the use of the string.lower() function as it is missing in Python 3. * Python 3 fix for the relax information printout. The try blocks seem to now operate slightly differently in Python 3. * Python 2 fixes - the Python 3 fixes for the ScientificPython module imports broke Python 2. * Python 3 - the cStringIO.StringIO import switches to io.StringIO if missing. * Python 3 - relative module paths are now used for the test suite runner. * Python 3 fixes for the version module for catching empty lists. * Python 3 fix - ensure an integer is actually an integer (division now converts ints to floats). * Python 3 - fixes for the renaming of the Queue package. * Python 3 - converted the last of the except error catching statements to be Python 2.4+ compatible. * Python 3 - removal of the use of the string.atoi and string.atof functions. These have been depreciated since Python 2.0! They have been replace by the int and float functions. * Python 3 - a number of fixes for running the ScientificPython modules in relax on Python 2 and 3. This includes relative imports, converting raise statements to function calls, removal of the use of many string module functions which do not exist in Python 3, etc. * Python 3 - modified some except statements to be Python 2.4+ compatible in a ScientificPython module. * Python 3 - converted some print statements to function calls in the ScientificPython modules. * Python 3 - fix for an os.chmod() call by using the stat module rather than the number 0775. The number 0755 is no longer valid in Python 3. * Python 3 - a pile of relative path fixes for many relax modules. * Python 3 - removed the use of the types module from generic_fns.sequence. The relax arg_check module is now being used instead. * Python 3 preparations - removed all of the string module functions which no longer exist in Python 3. These functions are part of the strings themselves now. * Improvements for the relax test suite synopsis for when the wxPython module is missing or broken. This is simply a printout improvement. * Python 3 preparations - removal of some unneeded xrange() calls. * Python 3 preparations - the data package now really does use the absolute path for its module imports. * Python 3 preparations - the data package now uses absolute imports for all its modules. * Python 3 preparations - eliminated the use of the types.ListType object. * Python 3 preparations - absolute module path fixes. * Python 3 preparations - support for both Python 2 __builtin__ and Python 3 builtins modules. * Python 3 preparations - absolute module path fix. * Python 3 preparations - more exception handling updates for all Python 2.4+ versions. * Python 3 preparation - all raising of RelaxErrors is now Python 2.4+ compatible. * Python 3 preparations - error handling is now Python 2 and 3 compatible in the relax_io module. * Python 3 preparations - converted the relax prompt/script interpreter to be Python 2 and 3 compatible. * Python 3 preparations - removed the use of the types.ClassType object. * Python 3 preparations - compatibility for both the Python 2 cPickle and Python 3 pickle modules. * Python 3 preparations - all usage of string.split() and string.strip() has been eliminated. * Removed the completely unused gui.components.conversion module. * Removed an unused import (which was breaking relax in Python 3). * Python 3 preparations - all os.popen3() instances in relax have been replaced by the subprocess module. * Python 3 preparations - eliminated the use of the os.popen3 function from the info module. * More exception handling changes to be Python 2.4+ compatible. * Python 3 preparations - exception handling fix to be Python 2.4+ compatible. * Python 3 conversions using 2to3. * Updated the Python 2 to 3 checklist document for the shifting of the 'relax' file to 'relax.py'. * Python 3 preparations - removed all usage of the xrange() in the generic_fns package as none are needed. * Python 3 preparation - eliminated the unneeded use of xrange(). * Python 3 preparation - the use of an absolute module path for import. * Python 3 preparations - the auto_analyses package is now fully Python 2 and 3 compatible. * Python 3 preparation - the auto_analyses package now uses absolute paths for the module imports. * Python 3 preparations - the use of the queue module in the status module is now compatible with 2 and 3. * Python 3 preparations - the GUI tests are now fully Python 2 and 3 compatible. * Python 3 preparations - the queue modules for both Python versions are now supported in the GUI tests. * Python 3 preparations - the test_suite.gui_tests package now uses absolute module path imports. * Python 3 preparations - the unit tests are now fully Python 2 and 3 compatible. * Python 3 preparation - all of the _generic_fns unit tests now use absolute module imports. * Python 3 preparations - all the _prompt unit tests now use absolute module imports. * Python 3 preparation - removed all xrange() calls from the unit tests, these are not needed. * Last Python 3 compatibility update for the system tests - they are now both Python 2 and 3 compatible. * Python 3 preparation - the test_suite.system_tests package now uses the absolute module path for imports. * Python 3 preparation - changed the import of SystemTestCase to use the absolute module path. * Removed all of the xrange() calls from the system tests as these are not necessary. This is in preparation for Python 3. * Some changes in preparation for Python 3. * Removed the 'force flag' text from the RelaxWarning messages output by the bruker.read user function. The force flag arguments of the generic_fns.mol_res_spin.name_spin() and generic_fns.mol_res_spin.set_spin_isotope() functions can now be set to None to suppress the text. * Fixes for the checks in the Mf.test_mf_auto_analysis() GUI test for the recent test suite data changes. * The CSA setting in the model-free auto GUI analysis now defaults to the '@N*' spin ID. Previously no spin ID was being used so that the protons where also having their CSA values set to that of the nitrogens. Now the execution checking code skips the proton CSA check. * Added star versions of the standard spin IDs to the spin ID GUI element (e.g. '@N*', '@H*'). * Fix for the comment on the 'Export' button in the BMRB export window. * Lots of editing of the model-free GUI section of the user manual. * Fix for the Relax_data.test_delete system test for the changes to the relax_data.read user function. * Fix for the Relax_data.test_read unit tests for the relax_data.read user function changes. * Fix to the Dasha system test needed for the changes to the relax_data.read user function. * Fix for the N_state_model.test_monte_carlo_sims due to the changed sphere.pdb test suite file. * Relaxation data is no longer loaded by relax_data.read if the values and errors are both None. * Modified the Mf.test_dauvergne_protocol system test to catch bug #20197 (https://gna.org/bugs/?20197). The sphere test data NE1 and HE1 data is now being used in this system test, triggering the bug. * Small change to the sphere model model-free test suite data. The trptophan indole data is now merged into the last residue (a glycine) to catch bug #20197 (https://gna.org/bugs/?20197). * The overfit_deselect() printouts for all specific analyses are now regularised and match the model-free printouts. * All overfit_deselect() methods now accept and use the verbose argument. * Printouts for the over-fitting deselection of spins are suppressed for the back-calculation of relaxation data. This affects the model-free Monte Carlo simulations, improving the output. * More improvements to the model-free over-fitting deselection printouts. * Improved the model-free overfitting deselection printouts prior to optimisation. Only a single message per spin is now given when the spin is deselected, minimising the amount of output. * Added a tryptophan NE1 data set to the sphere model-free model test data. This is in preparation to catch bug #20197 (https://gna.org/bugs/?20197). The scripts have also been updated for the newer relax designs. * Added the data_check Boolean argument to all of the specific analysis overfit_deselect() methods. This allows the unit tests to pass. Bugfixes: * Fix for the State.test_load_state_no_gui() GUI test - the data pipe bundle is now given. * Another attempt at fixing a nasty circular import in the relax GUI. * Fix for the relax user manual compilation. The full path for an icon in the graphics directory was being used. * API documentation fix - a circular import has been broken. This was only a circular import when the gui package is imported from an external program like Epydoc. * PDB format fix for the atom numbers of zero - the atom number should be consecutive and unique. * Big fixes for the running of the GUI tests. The GUI initialisation and termination, when not running in GUI mode, is now preformed outside of the Python unittest framework. The GUI was previously initialised in the setUp() method for all GUI tests, but for some reason this caused abrupt terminations of the GUI_test_runner.run() method. This run method, when finished, would not return to the Test_suite_runner but would cause the Python interpreter to silently die. The result was that the test suite summary would appear not to be printed out. * Python 2.3 and earlier fix for the unit test module for the float module. The 'tests' class variable was renamed so that the test suite does not think that it is a unit test. * Bug fix for the unit test runner - this restores functionality broken all the way back at r12100. The TEST_SUITE_ROOT constant no longer exists, so the check for it is now skipped. This allows tests to be run directly by the script. * Python 2.4 and earlier fixes for the Queuing module. The TaskQueue class from http://code.activestate.com/recipes/475160/ which was added to Python 2.5+ has been added to the compat module. This module is now used for all imports of the Queue class for all Python versions. * Test suite bug fix - the GUI tests now cleanly terminate the interpreter thread. This allows the test suite to terminate normally. With different Python versions on different operating systems, this could sometimes cause the test suite to freeze at the end, the final printout to be missing, or other strange behaviour. * Python 2.3- fixes - the Modelfree4 and Dasha system tests are now skipped if the subprocess module is missing. * Python 2.3 and earlier fix for the Test_relax_re.test_search() unit test. The unittest assertTrue() and assertFalse() methods do not exist, so assertEqual() is used instead. * Python 2.3 and earlier fix - the subprocess module is only imported when present. This is for the stereochem_analysis auto-analysis, but the import kills all of the auto-analyses. * Python 2.3 fix - eliminated the superfluous operator.itemgetter() call in the diffusion tensor code. * Python 2.4 and earlier fix for the GUI. Try-except-finally statement blocks are not supported, so these have to nested in two try statements. * Python 2.4 fix - the unit test runner no longer uses the builtin set() method. * Python 2.4 and lower fix for the pipe_loop() generator method. In these Python versions, a 'try' block with a 'finally' statement cannot contain 'yield' statements. This work around should be compatible with all Python versions. * Python 2.4 fixes - the ctypes module does not exist in Python versions 2.4 and lower. This is only used for some non-essential functionality on MS Windows, so now it is deactivated if not present. * Fix for bug #20213 (https://gna.org/bugs/?20213). The problem was that the generic_fns.sequence.generate() function was searching for a spin based on its spin ID generated from the spin name and not spin number. Hence two spins with the same name but different numbers were treated as the same spin, and only data from one was being generated via the copy() function. * Bug fix for wxPython 18.104.22.168 - the unsetting of the alpha mask for all bitmap images should now work. The wx.Bitmap.SetMaskColour(None) call does not work, so instead a wx.Colour() instance is passed in instead. * Fix for bug #20210 (https://gna.org/bugs/?20210) - relax now runs again without having wxPython installed. * Bug fix for the BMRB system test initialisation for when the bmrblib module is missing. The base __init__() method needs to be initialised first. * Fix for bug #20201 (https://gna.org/bugs/?20201). The bruker.read user function can now handle the situation where multiple spins per residue are already loaded prior to the user function call. The isotope value from the DC file is now translated to the element name and this is used as part of the spin ID string. * Fix for bug #20197 (https://gna.org/bugs/?20197) - the PyMOL and Molmol macro failure. The macro creation now only operates on spins called 'N' and assumes these are the backbone nitrogens. The sidechain tryptophan indole NH data will need to be detected and added later. * Fix for bug #20198 (https://gna.org/bugs/?20198) - the relaxation data back-calculation error. The problem was that the call to the model-free overfit_deselect() method was deselecting any spins lacking relaxation data. This should not happen, as no spins need to have relaxation data for this calculation.