TODO

Review CalcQat stuff for restraints - is this actually useful ?

memuse()  is non consistent and broken in places. Either remove compeltely 
(my preference) or consistentify.
CAn we globaly overload the new operator to count all the calls to new ?
probaly not. 


CHECK: Is the vdw parameters for HO atomtype in amber03aa wrong ???
CHECK: Is the vdw parameters for N3 atomtype in amber03aa wrong ???

CHECK charmm22.ff - examples/building/proteinhelix still throws errors !

Add make install hook to print a message about setting the PD_PARAM_PATH

====================
Unify UNITS !

Time:  PicoSeconds
InvTime:  PicoSeconds-1
Distance:  Angstrom
Mass:      AMU ?
Angle:      Radians
Charge:   ElectronicCharge
Energy:   kcal/mol

What about a series of functions:

 Units.cm( centimeter ), Units.Angstrom( angstroms )
 Units.nm( .. ) which return every thing in standard 

 Units.kcal( kcal/mol )
 Units.Joules ( 

 or 

 Units.Length.cm ( centimeter )

 Units.Energy.Joules( joules )

====== Python BackCalling ======

Can we subclass in python ? That would be Ueber awesum!

Monitors ?

MonteCarlo acceptance criterion ?


===== IO =======



PDB_In
	PDB_In is unintuitive - it doesnt load anything ! you ahve to do 
	sim.load first - fix somehow
	Wrappers ?

	sim.load() should load (ar attempt to) EVERYHTING - and complain if it cant.
	users can set filters to enable the auto-cleaning of PDBs
	Warning messages should point the user to these functions
	the default filter should be "everything" - but a real everything, not just
	all classes, so that you can read in an arbitrary thing in a PDB provided it
	is defined in the ffps.
	
	Default class for non-classed molecules
	File particle should be marked as taken when recognised. -> avoid multipl eimports.
	also warn about non-imported, unrecognised stuff. Advise user.
	
	
	Infile --> too PDB specific, move stuff out into the PDB_In

	Tutorial on how to make a new Inputfile, e.g. PSF	


	--> Trajectory player through workspace ...
	work in progress - a working version is now in place but rough edges remain, esp. the setting
	of periodic box changes is a little dirty/unfisnished.

	way to "save/load" a workspace/system
	PSF based setup/saving 
	Review & Document File Import, teach Mike and create a PSF importer


=======================

FileIn/Import .. etc.. TopologyIn/Import  
 	improove In vs Import etc..
 	

Trajectory:
	Trajectory/CoordinateSet loader/player through workspace ...
	way to "save/load" a workspace/system consistently through PSP/MIME/XTC/TRR/TRA 
	

====== Python =========

	print/cprint issue - syncronise python/C stream flushing
	stdout hooks ?

===== Workspace/System design ======

Workspace #1:
	Workspace should have a private =operator
	
Workspace #2: 
	WorkspaceCreator whci just clones and makes a new workspace.

Workspace #3:
	
	complete workspace ---> system updateing of positions
	
System #1:  
	Make solvating the system a SystemOperator.


AtomParamter/Particle Hiarchy - 
	split hiarchchy into 
	
	ParticleBase (= current Particle - cov arrays)
		\-- Particle (adds cov arrays)
		\-- AtomTypeParamter
		        \-- AtomParamter 
		        
	Particle.epot is a similar issue

	BUG: AtomPamameter parent strings are not set	
	
Bondorder manager

	change back the bondorder class to store **both** sides of the diagonal strip
	**including** the 0 in the middle... 
	
===== Error Handling/ Printing ====== 

Exceptions:
	replace remaining "return -1" error handling with "throw()" error handling
	Things still use the old error return int. 
	  - e.g. OutputTrajecotry.

Global verbosity system 

	- Loki - singleton ?
	- Arbitrary file handle siphening/stream siphening
	- OutputLevel will have to be a member of Object - and that almost everything should derive from object
	- Object/ClonableObject ??
	- Printf will be member function of Object
	
===== Satelite Modules =======

Proposal to create the 'Audition' and 'Experimental' satelite modules. The 
first to contain code that has been deemed desirable for PD to have, but is alpha-quality.
The 'experimenral' satelite module would contain algorithms which are completely unproven.

Proposal to move modules from the folder mods to the folder satelites to detinguish between
the two. Developers may add modules to /mods, but not /satelites.
Notsure:

	requirements checks - a mechanism such that each component
	(object) makes sure it is happy with the current setup.
	-- I think, related to this, that MoleculeBase needs some 
	sort of "ive been modified" serial number so that clients know, 
	and dont have to detect that their MoleculeBase has changed.



====== Forcefield =======


SECTION
	get rid of SECTION - repalce with CUSTOM PROPERTY service

====== Renamings / Cosmetics =========


Namespaces: 
	stuff to move into namespace
	namespace Core {
		System, Workspace, Primitives, etcc. 
	}
Renamings / Convetnions:
	- Private members - many still dont obey the m_BlahBlob convention 
	- Rename old stuff in FFPS


	
======= Other things ===========

	thread safety ?
	MPI ? How does it fit in ?
	

======= 

----------------- Pickers -----------------
	Pick_AND Pick_NOT etc are excellent, but...
	I have a function which needs a PickResidueBase and I have two deriving classes. I want to use a Pick_AND - i cant, becuase Pick_AND
	is an AtomPicker NOT a ResiduePicker.

	Solution - make copies of Pick_AND, NOT, OR etc for  PickResidueBase called PickRes_AND etc. But then what if we then have a new base class for another subset of pickers? We would have to duplicate all of them again!!



-------------------------------------------------------------------------
--------------------<  DONE  >-------------------------------------------

Cleaned error reporting in ffps reader.

DONE RENAMINGS:

	External:

	- All forcefields should migrate to FF_Xxxxx_YYZZ
	- All out output trajectories should be OutTra_XXXX (including PDB_Out and Traj_Out )
	Internal:

	nAtoms()  !
- Old "TRA" --> Bristol Trajecotry Format (btf)
Renamings:

BondedForcefield --> FF_Bonded
CustomForcefield --> FF_Custom

Merge PickedMinimise vs Minimise

CHARMM 
	CHARMM reader - for import of Charmm and OPLS definitions from 
	the standard charmm distribution
Forcefields
	Other forcefields:
	AMBER older versions
	CHARMM 
	OPLS
	Gromos
	TIPXX, SPC wter models	

 PD_LIB_PATH --> PD_PARAM_PATH

Build system - configure/autoconf/automake
Installer into python lib directory

Actually us the WorkSpace operator class as a base for everything that
takes a Workspace reference. Unify how to get to the stored pointer
(getWSpace() ) or just wspace() maybe (i prefer that)

forcefields should initialize with wspace - this will make the setup 
function less important and makes things much easier when setting up
picked forcefields in particular.

ff.info() shouldnt be const - it must call setup()
although if we get rid of setup (see point above) then this problem 
goes away .. :D

do that atom.xyz fix (such that molecule::atom[] points to wspace.cur in workspace)

RotatableBondlist should be a Null class by default - it takes a long time to set up,
probably lots of memory and is hardly ever used.

Integrated Maths:dvector into SWIG/Python environment

Forcefield "LIB" path environment variable

fix long-range bondorder (sparse) in the new neighborlists !

Gromacs outtra, xtc, and other formats such that g_tools can be utilised and interfaced with

fix module generation - this will important for removing half backed
crap and experimental research-specific stuff into modules!. Compiling all
these modules will involve the build system stuff (see above);

Fixed virial

Fixed physical constants




======== FUTURE =========
---- Jons BIG todo list
1) Port the ProSup structure superimposition code to PD
3) Improved sequence alignment ->
   we still dont use that Blosum Matrix - not hard to implement. 
   Implement local squence alignment algorthms too? Multiple alignment? 
   Or simply importers to obtain data from externam ClustalW etc. alignment engines in FASTA format etc.
4) DSSP / STRIDE port to allow structural annotation