Rx is found in the files `rx.h' and `rx.c'.  It is an upward
compatable replacement for GNU regex.  If you want to try GNU Rx in
your own application, the documentation for GNU _regex_ will tell you
how.

Notes:

* This release of Rx is not quite ready for use in GNU Emacs.

* When initializing a variable of type `struct re_pattern_buf',
  use `bzero' or an equivelent function first.  Some older code
  might try to work by assigning 0 to specific fields defined in 
  GNU regex.  All of those fields still exist, but new fields have
  been added (and more may be in the future).

* Rx keeps a cache of information about patterns.  The size of this cache 
  may effect performance.  The default size should be large enough for
  most applications.  However, for a program like `sed', there is no reason
  to keep the cache small.  Such programs might want to assign a large 
  number (say, 10000) to the global variable `rx_cache_bound' before
  calling other rx functions.

  Finer shades of allocation control are possible, but there is no 
  documentation to describe how, yet.

* Like GNU regex, Rx can return the positions of parenthesized subexpressions
  in a match (`registers').  Rx adds a new field to an `struct re_pattern_buf'
  called `syntax_parens'.  That field may be used to tell Rx to ignore some
  parentheses; i.e.: to treat them as purely syntactic and not return
  the location of the subexpression they surround.  Using this field is 
  optional, but may speed up matches.  Full documentation for this feature
  is lacking, but adventurous users can find an example of its use in `sed.c'
  and can have questions answered by writing to the bug address.