FUTURE GOALS
------------
Listed in order of priority :-

* Continue eliminating bugs.  If you find one, make sure you report it
  as maybe no-one else will!

* Improvements to library performance, based on real usage as reported
  by users.  If you find your code is running slower than Mawk/Gawk,
  why not forward me an example.  I have no idea how fast Awka is 
  compared to Tawk.

  There are several benchmark tests in the benchmark directory (where 
  else?) - more & better ones would be useful.
  
  At present regular expressions are quite slow.  This is a complex issue
  to solve, but I'm working on a solution as & when I get time to do so.

* Useful debug warning messages to tell you if you've got common bugs in
  your AWK program, such as the old typo error with variable names.
  
  The framework data structures for these are now present in the translator;
  I just need time to implement them.

* The ability to define your own C functions to be called from the
  AWK program code.  I have a draft version of this just about ready to
  go, however I want to get a stable Awka version out first, as this would
  be a dramatic change that'll probably introduce a host of new bugs.

* AwkaTk (only if a requirement exists, although it would be interesting).
  Perhaps AwkaQt?  AwkaGtk+?  Potentially these could be implemented via
  the C function idea above.

* Gawk extensions to regular expressions and printf format specifiers.
  Also Gawk internet features.

* A document detailing the library API, useful for those who want to 
  work directly with the generated C code.  Better documentation all
  round would be good.

* Make Awka thread-safe, or potentially allow compilation of a separate, 
  thread-safe version (it would run slower).  To be honest, I'll probably
  never do this.

* Extensions to the AWK language.  I'm thinking most about introducing
  structures in some way.  Not only is this a major change, it may
  also upset many people (not that they'd have to use it!), so I'm
  not really sure about this yet.  I know Tawk has done something
  along these lines, but I couldn't promise a similar implementation.
  I'm also thinking about real multidimensional arrays (ie a[1][2]).

* More efficient, complete and elegant C language generation, including
  nice things like generation of commented code, and preservation of
  AWK program comments.

* More platforms, maybe including a Windoze DLL compile of the library.
  As both Cygnus and DJGPP are supported, I'm not currently too concerned
  about doing a native Windows version.

* Creating translators for AWK to Delphi (Pascal), Visual Basic and any 
  other language where there seems to be a demand for AWK-like functionality.
  For this I will need active involvement from people with expertise in
  these languages/environments.  I'd probably also have to do a Windoze 
  DLL compile 8-(.

