Perltidy TODO List
    This is a partial "wish-list" of features to add and things to do.

  Perltidy not idempotent
    Perltidy does a single pass but iterates locally to try to satisfy the
    various constraints and formatting parameters, but in some cases it can
    take a second pass to reach a stable format, particularly when major
    format style changes are made. It would be useful to have a flag to
    force a second pass, say --passes=2.

    For now, a workaround for a single file is

      perltidy file.pl -st | perltidy -o file.tdy

  The starting indentation level is not correctly guessed when tabs are used
    When formatting a snippet which does not begin at zero indentation,
    perltidy must guess the starting indentation. The -sil flag can be used
    to specify this, but this is tedious. In general we cannot know the
    indentation and can only guess, but we can at least require that if
    perltidy is run twice on the same snippet, with the same parameters,
    that the indentation be correctly guessed on the second pass. Currently,
    this is only true if there are no tabs in the leading whitespace. A
    patch has been submitted to fix this but it needs a little work still.

  -b does not work through the Tidy module
    The use of -b in the following does not work:

       use Perl::Tidy ();
       use File::Spec;

       my $file = File::Spec->catfile( $dir,
          $filename );

       Perl::Tidy::perltidy(
          source     => $file,
          argv        => '-b',
          perltidyrc => $perltidyrc,
       );

    A temporary workaround is:

       Perl::Tidy::perltidy(
          argv        => "-b $file",
          perltidyrc => $perltidyrc,
       );

  Improved Vertical Alignment
    There are still many opportunities for improving vertical alignment.

  Documentation
    A FAQ is needed to explain some of the more subtle formatting issues,
    and to give examples of different styles.

    The -formatter callback object documentation is incomplete.

  HTML writer
    The HTML writer does not colorize interpolated identifiers in here
    documents or quoted strings. The tokenizer outputs interpolation
    information for here docs; it still needs to be updated to do this for
    multi-line quotes. Then code needs to be written to scan for and markup
    identifiers.

Things which have been suggested but will not be done
  Recursive file processing
    A -r flag might be nice, but this is best handled by an exterior shell
    script.