
			CML, XML and costwish
			---------------------

			   Things to do

Since these systems are intended to be extensible there will always be
a large list of things that can be done.  This list represents high priority
TODOs that have occurred to me over the last few months.  They are separated
into the various systems - their inclusion here is no guarantee that I am 
working on them!

XML DTD and general
-------------------

1. Many data have ranges (e.g.  27.3(2), 45.1+-1.2, 3.0-7.2, " >1.2"). A new
   attribute for X.VAR and X.ARR could support this.
   -- The new attrubute FUZZY is being developed to cater for this --

2. X.SYMB (mathematics and symbolic variables) is completely unexplored.  It's
   fairly urgently needed where quantities depend on one another.

3. There should be a system for supporting large numeric data sets (HDF,
   or Scott Nelson's model/ MIME proposal.

4. XML needs a CLASS attribute.  This may have to wait until HTML-WG decides
   what they are doing about this.  I might introduce a USERCLASS attribute
   which would serve some of the purpose and mutate later.

5. is XMLTYPE still needed? (I don't use it).
   -- deleted --

6. 2- and 3- geometry should be possible (using X.VAR or X.ARR?).  E.g.
   <X.VAR TYPE=CIRCLE>1.2 3.4</X.VAR>.  POINT2 POINT3 LINE2 LINE3 CIRCLE
   SPHERE PLANE POLYGON.  These were all worked out in Democritos
   (http://www.dl.ac.uk/CBMT/)
   -- allowed under X.ARR, but semantics not yet crystallised --

7. Internal identifiers (e.g. NAME="#REF:REZP94") are not formalised.


MOL DTD
-------

1. Reactions are not catered for.  Now that MOL.DTD supports multiple 
   molecules and pointers, reactions can be defined in terms of 
   self-contained molecules in the file.

2. Generic molecules (searches, markush, etc.) are not catered for.  

3. C.MAP (mapping from one molecule to another is not explored yet.  Is it
   a case of a more general relationship required in X.VAR?

4. Sequences are not yet well managed.  C.SEQ and C.CHAIN is overkill.
   Maybe simply C.CHAIN*.  What about ligands to chains?

5. There will be some more BUILTINs for C.AT and C.BO.  

6. Dummy atoms and generic atoms are not well managed.

7. Can C.AT and C.BO have columns with lists of values?  If so, does it 
   need X.LIST?

8. SERNO, SERID for atoms are not yet consistent.

9. 

Costwish
--------

These are DTD-independent things (or very nearly) that need to be done for
all applications.

1. Porting costwishexe to other platforms
   -- wide range of UNIX --  PC has to wait for more tcl work

2. HELP.
   -- there is a tutorial now --

3. CCI if possible (at present uses html_library.tcl for its hypertext).
   It needs a fluent link to browsers for external URLs.

4. Rewriting in itcl.

5. Writing the load routines in C (they are quite slow for large files).

6. User-selection of nodes and their properties.

7. LaTeX output.

8. Bug in CML output (too many newlines).

9. Subtrees alternatively displayed as TOCs rather than frames 
   (recursive windows)
   -- done --

10. 

CMLcost
-------

This relates specifically to the postprocessing of XML and MOL.

1. Units added uniformly to widgets.

2. 'Alternative units' offered consistently.

3. Scrolling spreadsheet.

4. Truncation (or wrap) of spreadsheet titles.

5. FUZZY needs a postprocessor

6. So does SEQ

