| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Pressing any key before the player thread set mode to PLAYER_INITIALIZED
lead to a song skip and spawning of another player thread. The two
player threads were operating on the same variables/memory area,
resulting in a crash.
Adding a new player state (PLAYER_STARTING) and setting it _before_
invoking pthread_start should resolve this, 'cause the "if
PLAYER_FREED-branch" is not taken any more, ergo no second player thread
is started.
|
|
Under certain circumstances (fast station switching e.g.) a new playlist
was not fetched, because the player thread returned 0x1 (== libao error)
and the main thread NULLed curStation.
|
|
|
|
|
|
Time is now counted in milliseconds, not aac frames. Calculations seem
to be inaccurate sometimes (about +-2 seconds).
|
|
Now libfaad and/or libmad are used for playback. There's currently no
remaining time displayed for mp3 playback.
|
|
|
|
Update copyright notices, add my email address, but don't change
libwardrobe (I'm not maintaining it anymore)
|
|
This fixes those my-favorite-song-suddenly-aborted-"bugs", as well as
the pause mode, which was more like a stop mode, because pandora
*always* aborted the transfer after a few minutes.
|
|
Some code restructuring was made, too:
- Move UI functions to ui.c and ui_act.c
- Change default keys for loving and banning to + and -
|
|
It's not necessary to re-compute the scale factor as it will be the same every
time.
|
|
Could be tuned...
|
|
Smarter solution than commit 8421cfd9886ec5ee19cde18b48853c2f9ea60c29.
|
|
|
|
Now unneeded header, linebreaks
|
|
This has two advantages:
1) We can pass the correct amount of bytes to NeAACDecDecode (). This should
fix some decoding error messages. And
2) We can show the length of the song and the remaining (or already played
time) to the user.
Cool, eh?
|
