- Correcting the Input (TreeTagger Output) to be always conformed (use a method defined in YaTeA - todo)

- Add POD directive =encoding in each file

- Delete the Options_EN and Options_FR. 

  Replace it by the loading of the file share/YaTeA/yatea.conf

  (function LoadOptions)

  use of Config::General for config management

- in ParsingPatterns_EN, take into account to the tag NP (?)

- Add a prefix while making RESULTS directory

- Other method to load config data (rather than change ROOT at the
  installation step) :


sub openunicode {
    my ($rfh, @path) = @_;
    my $f;
    unless (defined $$rfh) {
	for my $d (@INC) {
	    use File::Spec;
	    $f = File::Spec->catfile($d, "unicore", @path);
	    last if open($$rfh, $f);
	    undef $f;
	}
	croak __PACKAGE__, ": failed to find ",
              File::Spec->catfile(@path), " in @INC"
	    unless defined $f;
    }
    return $f;
}

- Move bin/.config information in yatea.conf (?)

- Write a more explicit message when corpus file does not exist

- Move the Option MESSAGE_DISPLAY in the yatea.rc

- Decription of the terminology format

- rewrite the method
  Lingua::YaTeA::ForbiddenStructureAny::setSplitAfter and its calls

Check the XML conformity, especially < and > and &


- Check the computing of the offset. it is assumed that every token
  are preceded and follow by a space character. But it is not always
  the case (e.g. commas)

- Check that special XML characters are well encoded as XML entities

- Homogenize the output of the file and file handler (print / debug -> file handler, printUnparseable -> file name)

- remove the dtd line in the output (to check ?)

- UTF-8

- tests : loading a corpus
	  loading  a testified term list
          chunking
          progressively parsing
	  displaying results
	  process English corpus
          process French corpus (tagged with TreeTagger)
          process French corpus (tagged with Flemm)
	  process English corpus with testified terms

- to check that Edge.pm is useful

- review WordFromcorpus and LexiconItem (the word form is concated and then split !)

- integrate message display in submodule (for instance TestifiedTermSet)