Log Message:
Updated to version 0.9.6

Changes:

- mkvextract: bug fix: The track extraction was creating the
  output file twice if the Matroska file contained a copy of the
  track headers. This resulted in the first extracted file being
  overwritten at the end of extraction.

- mmg: bug fix: If the file title is read from an input file, not
  modified by the user and that input file is removed again then the
  file title will be unset.

- mkvmerge: enhancement: Converted the raw FLAC reader to use
  another interface to the FLAC libraries. This results in a speedup
  of up to 50%. Thanks to Josh Coalson for telling me about its
  existence.

- mkvmerge: new feature: Added two warnings. One about invalid
  track IDs that were used on the command line but that don't
  correspond to an available track in a file and one if no track
  will be copied from a source file. Both warnings hint at bad
  command line arguments.

- mkvmerge: Only write the segment duration as a 64bit float if
  there is no video track present. This way users won't have to
  update their DirectShow filter/apps for most files. Only
  audio-only files need this precision anyway.

- mkvmerge: Changed the Ogg/OGM reader to use the stream number
  and not its serial number as the track ID (meaning the track IDs
  will be 0, 1, 2... etc. instead of the random numbers oggenc uses
  as the serial numbers).

- mkvmerge: new feature: The CUE sheet parser now accepts INDEX
  lines with indices from 00 up to 99 and implements the Red Book
  specification for audio CDs that way. Patch by Vegard Pettersen
  <vegard_p at broadpark adot no>.

- mkvmerge, mkvextract: bug fix: ASS was handled like SSA which is
  not correct in each case, especially when extracting it.

- mkvextract: bug fix: The WAV writer was not endian safe.

- mkvmerge: bug fix: The charset was not set correctly on Solaris.

- mkvmerge: bug fix: mkvmerge crashed when reading Matroska files
  that contain an empty tag list.

- mkvmerge: bug fix: Some Matroska files that e.g. have had their
  timecodes offset with the Matroska Stream Editor or other means
  may contain timecodes that caused mkvmerge to print a warning
  about "timecode < last_timecode". A new fix implements a
  workaround and a warning message with a proper explanation for
  this case.

- mkvmerge: bug fix: Older Matroska files containing chapters
  caused mkvmerge to abort muxing.

- mkvmerge: bug fix: mkvmerge was only copying the last tag of a
  list of tags applying to a track from a Matroska file.

- mkvmerge: bug fix: mkvmerge will show a nice warning if the
  entries in a SRT have non-continuous timestamps. It'll also sort
  the entries by their start timestamp instead of throwing the
  generic "timecode < previous timecode" warning.

- mmg: bug fix: The 'Matroska file analysis' window that occurs
  when reading chapters from a Matroska file did not disappear if it
  was minimized when the process finished.

- mkvmerge: new feature: Added a new parameter
  --aspect-ratio-factor.

- mkvinfo: bug fix: Strings from chapters and tags were shown in
  UTF-8 instead of the local charset. This bug was introduced around
  2004-08-28.

- mkvmerge: bug fix: Not all chapter elements were copied
  correctly from a source Matroska file.

- mkvextract: Sped up the extraction of attachments, chapters,
  cuesheets and tags by using the seek head information and not
  parsing the full file each time.

- mkvmerge: bug fix: The Matroska reader was not handling very big
  cluster timecodes correctly. Those can occur when the timecode
  scale factor is very small.

- mkvmerge: bug fix: Empty clusters in Matroska files no longer
  make mkvmerge think that file has been read completely.

- mkvmerge: new feature: Added support for MP2 (and maybe MP3)
  audio in MP4 containers.

- mkvmerge: new feature: The chapter and tag parsers accept XML
  element attributes instead of sub-elements for those sub-elements
  that only contain data. Example for a "simple tag":
  <Simple Name="ARTIST" String="Tori Amos"/>

- mkvmerge: bug fix: The automatic MIME type detection based on
  the file name extension was using the file name extension as the
  MIME type.

- mkvmerge, mkvinfo: new feature: Added the four new PixelCrop
  elements.

- mkvmerge, mkvextract, mkvinfo: new feature: Added
  'TargetTypeValue' as a supported tagging element.

- mkvmerge, mkvextract, mkvinfo: Complete rewrite of the chapter
  and tag parsing and output functions. Additions will be much
  easier now.

- mkvmerge, mkvextract, mkvinfo: feature removed: Dropped support
  for the very old and deprecated tagging system. No one used it
  anyway.

- mkvmerge: new feature: Allow the use of two-letter ISO639-1
  country codes in for the '--language' parameter. Those will be
  converted to the corresponding ISO639-2 language code
  automatically.

- mkvmerge, mkvinfo, mkvextract: new feature: Added support for
  the 'TargetType' tag element

- mkvmerge: bug fix: The MP3 handling was broken on weird and rare
  occasions when reading MP3 from a Matroska file.

- mkvmerge: bug fix: Removed a bogus warning about an attachment's
  MIME type having been given more than once.