TODO:
- store drive features in a database
- try http://www.ime.usp.br/~pjssilva/secure-cdparanoia.py and see if it
  is better at handling some bad cd's
- .cue file:
  - add version to morituri comment
  - add DATE and CATALOG and PERFORMER and TITLE to top of cue
  - add TITLE AND PERFORMER to each track
- change format to be %2d - %performer by default
- at least mention the data track somewhere in the log
- check pregaps more than once, to see if results are consistent, or with
  different methods
- handle errors on cdrdao spawning (for example, not having cdrecorder,
  or not putting the disk in)
- check if it's simple to listen to each track in a multitrack completing
- save trms to a pickle, after finishing each track
- rip the data session
- add AccurateRip validation for ripped images to rip command
- add GUI
- write moovida/xbmc plugin
- cache results of MusicBrainz lookups
- on ana, Goldfrapp tells me I have offset 0!
- don't keep short HTOA's if their peak level is low
  (see Pixies Planet of Sound single)
- lossy encoding
- consider basing ripping progress not only on read (reaches 100% before
  writes are done) or writes (very bursty in cdparanoia) but a combo of the
  two, each counting for half.
- if disk not found in accuraterip, it doesn't mean that it's not accurate
- discs we should rip:
	LCD soundsystem disc 2 (data track)
- check if cdda2wav or icedax analyze pregaps correctly
- Zita Swoon anthology cd 1 shows track 8 rip NOT accurate, but checksums match
- burn ripped images
- use a temp dir, until the whole rip is good don't move it, so we easily find
  half done rips
- primal scream, track 5, it rips/checks/rips/checks/encodes/checks and then complains read and verify failed.  why did it encode ?
- Musicbrainz disc id error:
  Compare http://www.musicbrainz.org/cdtoc/MAj3xXf6QMy7G.BIFOyHyq4MySE-
  with http://www.musicbrainz.org/cdtoc/USC1utCZbTLZy80aHvQzJw4FASk-
  Almost same, but second is 2 seconds longer on last track, suggesting it
  was calculated wrong (150 frame offset done wrong ?) Can't find it in
  edit history though
  Write an example document with this cd as an example explaining offsets
  and id calculations
- possibly figure out how to name albums with credited artist; look at gorky and spiritualized electric mainline
- when shortening file name then reripping, it rerips since it doesn't know
  the shortened name; see sufjan stevens
- if a disc output dir is already there, see if it was properly finished;
  complain if it was, to not overwrite
- if multiple releases with different artist match the disc id, stop and
  let user continue by choosing one
- artist-credit-phrase fabricated by musicbrainzngs only looks at name, not at artist-credit->name (see e.g. Gorky)
- getting cache results should depend on same drive/offset
- do some character mangling so trail of dead is not in a hidden dir
- fix %r for normal case release name
- decide whether output-dir should be part of the relative filenames of things;
  right now it is; maybe split in to base and output ?
- rip task should abort on task 4 if checksums don't match
- retry cdrdao a few times when it had to load the tray