BUGS:
	if msgfmt is missing, the autoconf script still tries to rebuild the .gmo
	  files, in which it fails (workaround: ./configure --disable-nls)
	two feeze/thaws series in a row when using autoscroll slows things
	  down quite a bit
	the size of some windows is too small sometimes
	using electric fences, the gtk_set_locale() calls causes a segmentation
	  fault in gtk_init, I don't know who's to blame
	verify doesn't support the exclude pattern
	some cd-rws require a media reload before blanking; gcombust doesn't
	  support this
	the multisession interface is a little bit confusing
	it seems like the #ifdef fix for different gtk+ versions has to be
	  a dynamic fix instead as rpm people use it with different versions
	  of gtk+ than it was compiled for
	check that boot file exists before starting
	if burning an audio from and to the same device, cdrecord gets passed a (null)
	  value for some variables - it is of course a stupid thing to do, but it still
	  shouldn't generate (null) values

TODO: (incomplete list, not strict order)
	remove dead code from old make audio command routines
	try using mmap() instead of read() in image-copy operation - might not be
	  possible, but wouldn't eat cache on Linux (I think)
	make prefs for buffersize in cd-2-image copy
	better verify operation (not using diff)
	mount image and view it
	make disable tabs optional
	display available devices
	make the boot image creation more intuitive
	cd-to-cd audio copying
	mp3->wav
	mp3->cd-audio (mpg123 | cdrecord)
	improve all save/load related things
	version check of programs
	bin/cue (no support from cdrecord at the moment, at least not for vcd tracks)
	a few default setups for different tasks (quick selection of parameters)
	should be possible to move around files/dirs and change the
	    image dir structure arbitrarely (maybe use a tree in the display
	    instead of the clist)
	check the strdup/malloc/free's, it leaks memory in a few places (fixed most)
	    (calls to choose_path)
	there's some mixing of g_free/g_malloc and free/malloc, I later learned this
	  might be bad
	more doc
	gnome compliant
	remove a few static buffers
	make gcc not complain with -pedantic
	autodetection of burners should probably be done with cdrecord -scanbus
	    instead
	maybe make it possible to burn ext2 filesystems
	print out output of -inq also
