This is the BUGS file for the Cats library

* The LANG variable can be either xx-YY or xx.  The Cats library can
  read either.  Steffen Kaiser also makes this suggestion:
  >I made some sort of thoughts about this problem in the past for various
  >projects, the only somewhat suiteable method I found was this:
  >+ Imagine a function like this: int fileopen(char *fnam, int mode)
  >+ By default this functions simply calls open(fnam, mode), but on failure
  >the path of 'fnam' is extracted, then opened, and on success the first
  >line is read from it, then it is closed and the filename of fnam is
  >appended and the open() call is retried.
  >
  >This method allows the following scenario:
  >%NLS%\en-US
  >%NLS%\en-UK
  >%NLS%\en\LC_MESSAGES
  >
  >where "en-US" and "en-UK" are files containing:
  >%NLS%\en
  >
  >and "LC_MESSAGES" is the file the messages are located in.
  >
  >"%NLS%" should be the absolute pathname, though.
  >
  >This method does not simulate symlinks, but offers some options with
  >little overhead.
  >
  >For INI files I had implemented a similiar idea, where there is a general
  >token, e.g. "#!chain to file = <filename>", and whenever the file parser
  >hits such line, the currently opened file is closed and <filename> is
  >opened, so the file parser can go on without any hassle, thus, minimum
  >overhead, if the token is the very first line, however.


FIXED:

This was fixed in release 2.0:
* May not be very efficient.  Specifically, the database will
  be re-sorted when a new value is inserted.  Should only re-sort the
  database as needed when fetching values, not when inserting.
