##=============================================================================
## 
## This file is part of BibTool.
## It is distributed under the GNU General Public License.
## See the file COPYING for details.
## 
## (c) 1996-2004 Gerd Neugebauer
## 
## Net: gene@gerd-neugebauer.de
## 
##*****************************************************************************


This file contains ideas for improvements.

Most of the  items are mere  key words and  may not be meaningful  for
others.  I try to  keep the items in  a  order form highest to  lowest
priority.  However, I change the order from time to time and hopefully
items will disappear at some time.


General ideas:

    *   Make a C library of routines for manipulating BibTeX files.
	(partially done. The main emphasis went into the documentation.
	 Additional cleaning needed)

	- Use the lib and provide a interface to various programming
	  languages. The first candidate would be C. Other candidates
	  are C++, Tcl, Perl, Guile.
	- Use the lib and provide an SQL interface (see below).
	- Use the lib to write a better BibTeX.

    *	Add a SQL like extraction language (partially exists).
	Maybe this is the time to completely redesign the resource
	language. I think I will make a major revision or a seperate
	program/library for this.

    *	Use Unicode UTF8 as character set. This goes beyond BibTeX and
	brings in a new quality.

Special ideas:

    *	Provide a better installation routine. Especially for BibTcl a
	wish script might be the best solution since wish has to be
	installed before BibTcl can be made.

    *   The different keys: Sortkey/OldKey deserve a clearer handling.
	Maybe some bugs are hidden here.

    *	Maybe an index/several indices should be used to speed up
	access to records. Maybe a real database engine should be used for
	the underlying mechanism.

    *	Support for the extended  key notation file::key. This is said
	to be used in BibTeX1.0.

    *	Support the special fields @ALIAS and @MODIFY as described by
	Oren Patashnik in TUGboat. (partially done) Each function has
	to be checked to decide if a refinement of the specification
	is necessary

    *   Implement a garbage collector for the symbol table to avoid
	memory leaks when used from C/C++/Tcl/Perl/...

    *	An escape character in strings might be missing.

    *	Integrate field rewriting into the key generation algorithm.
	Allow arbitrary fields to be constructed like keys.
	Maybe this is a step towards a general manipulation language.
	I should keep this in mind for major revision 3.
	(The reverse is done. Key formats are evaluated when rewriting)

    *	General formatting language (Maybe a replacement for BibTeX)
	Other applications: extract all keys, prettyprint entries in
	pure ASCII (HTML), etc.
	The main problem is to make a decision for the scripting language.
	I think I should use something existing and not roll a new one.
	(Tcl, Perl, or Guile are likely candidates)

    *   Use a tree representation in field rewriting instead of parsing and
	evaluating the format each time (speed issue).

    *   Do a better job when sorting. Especially for international
	bibliographies the sorting may not be adequate. Learn something from
	xindy?

    *   Analyze bst files to find record types and entries used.
	(partially exists but is not integrated yet)

    *	Provide a specification for the statistics function. Maybe SQL
	serves for this purpose.

    *   Use the kpathsea library to find files.
	It has been done for Unix-like operating systems. For others it's
	still missing.

    *   Selection of items according to regexp match on fields.
	This has to be generalized to allow full power of a 
	first order language.

    *   Make the program 16-bit clean. This means UniCode!

    *   Eliminate compiled in defaults completely.
	Use a library directory to load initializations.

    *	Integrate add/delete.field into the rewriting mechanism.

    *	Automatic string detection.
	The reverse of string expansion. Replace common values by new strings.
	This can partially be done with field rewriting!

    *	Respect crossref at selection

    *	Support \cite in fields.

    *	Documentation: improve tutorial part. (Maybe a book?)

    *	Documentation in texinfo format?

    *	Documentation in HTML format?

    *	UN*X man pages?


   *** Write a graphical front end for BibTool

    -   Maybe not a simple task. BibTool is already too powerful.
	(first experiments are not very promising)

   ***	Write a replacement for BibTeX :-)
	Maybe I will wait until BibTeX 1.0 is out.

    -	Better styles
	in a readable programming language.
	This is a major problem. Which language should I use?
	I am working towards making a library. Thus it should be easy to
	integrate BibTool into any language.
    -	Object oriented approach.
	e.g. for proceedings containing articles.
	@proceeding{abc,
		editor = "...",
		booktitle = "...",
		title = "...",
	    @inprocedings{1,
		author = "...",
		title = "..."
	    }
	}

    -	Drop limitations. E.g.
	- size of a field.
	- order of crossreferenced records

    -   Support multi-language bibliographies.

    -   Backward compatibility: It should be possible to use the new program
	as a replacement for the existing BibTeX without modification of the
	BibTeX database.
	The styles should be usable or an automatic translation into a new
	format should be provided.



If you have read until here and you want to take one of the tasks then
contact me.

Gerd Neugebauer
gerd@informatik.uni-koblenz.de