[FFmpeg-soc] [soc] libavsequencer [PATCH 03/08] Order list public API header file.

[Sebastian Vater]

Vitor Sessak a écrit :
> On 07/13/2010 09:49 PM, Sebastian Vater wrote:
>>
>
> Is it the list of the rows in chronological order? Is it the list of
> notes in alphabetical order? Also, if I understand well, it is an
> array of linked lists. Why?

Because the user can insert new rows, delete old ones, rearrange them
when editing.

>
>>>>      /** Number of order list data used for this channel.  */
>>>>      uint16_t orders;
>>>>
>>>>      /** Number of order list data entries to use for this
>>>> channel.  */
>>>>      uint16_t length;
>>>>
>>>>      /** Repeat start order list data number for this channel.  */
>>>>      uint16_t rep_start;
>>>>
>>>>      /** Volume level for this channel (defaults to 255).  */
>>>>      uint8_t volume;
>>>> #define AVSEQ_ORDER_LIST_VOLUME 255
>>>>
>>>>      /** Sub-volume level for this channel. This is basically channel
>>>>         volume divided by 256, but the sub-volume doesn't account
>>>>         into actual mixer output (defaults 0).  */
>>>>      uint8_t sub_volume;
>>>> #define AVSEQ_ORDER_LIST_SUB_VOLUME 0
>>>>
>>>>      /** Stereo track panning level for this channel (defaults to
>>>>         -128 = central stereo track panning).  */
>>>>      int8_t track_panning;
>>>> #define AVSEQ_ORDER_LIST_TRACK_PAN  -128
>>>>
>>>>      /** Stereo track sub-panning level for this channel. This is
>>>>         basically track panning divided by 256, but the sub-panning
>>>>         doesn't account into actual mixer output (defaults 0).  */
>>>>      uint8_t track_sub_panning;
>>>> #define AVSEQ_ORDER_LIST_TRACK_SUB_PAN  -128
>>>>
>>>>      /** Stereo panning level for this channel (defaults to
>>>>         -128 = central stereo panning).  */
>>>>      int8_t channel_panning;
>>>> #define AVSEQ_ORDER_LIST_PANNING    -128
>>>>
>>>>      /** Stereo sub-panning level for this channel. This is
>>>>         basically channel panning divided by 256, but the sub-panning
>>>>         doesn't account into actual mixer output (defaults 0).  */
>>>>      uint8_t channel_sub_panning;
>>>> #define AVSEQ_ORDER_LIST_SUB_PANNING    0
>>>
>>>>      /** Compatibility flags for playback. There are rare cases
>>>>         where order handling can not be mapped into internal
>>>>         playback engine and have to be handled specially. For
>>>>         each order list which needs this, this will define new
>>>>         flags which tag the player to handle it to that special
>>>>         way.  */
>>>>      uint8_t compat_flags;
>>>
>>> I suppose this is not unused ATM?
>>
>> Fixed.
>>
>>>
>>>>      /** Order list playback flags. Some sequencers feature
>>>>         surround panning or allow initial muting. which has to
>>>>         be taken care specially in the internal playback engine.
>>>>         Also sequencers differ in how they handle slides.  */
>>>>      uint8_t flags;
>>>> #define AVSEQ_ORDER_LIST_FLAG_CHANNEL_SURROUND  0x01 ///<  Initial
>>>> channel surround instead of stereo panning
>>>> #define AVSEQ_ORDER_LIST_FLAG_TRACK_SURROUND    0x02 ///<  Initial
>>>> track surround instead of stereo panning
>>>> #define AVSEQ_ORDER_LIST_FLAG_MUTED             0x04 ///<  Initial
>>>> muted channel
>>>>
>>>> } AVSequencerOrderList;
>>>>
>>>> /**
>>>>   * Song order list data structure, this contains actual order list
>>>> data.
>>>>   * New fields can be added to the end with minor version bumps.
>>>>   * Removal, reordering and changes to existing fields require a major
>>>>   * version bump.
>>>>   */
>>>> typedef struct AVSequencerOrderData {
>>>>      /** AVSequencerTrack pointer to track which should be played.  */
>>>>      AVSequencerTrack *track;
>>>
>>>>      /** Next order list data pointer if seeking forward one
>>>> frame.  */
>>>>      AVSequencerOrderData *next_pos;
>>>>
>>>>      /** Previous order list data pointer if seeking backward one
>>>>         frame.  */
>>>>      AVSequencerOrderData *prev_pos;
>>>
>>> These do not look to belong to the BSS.
>>
>> They are used to allow the composer to define alternative order
>> sequences in case of forward and backward seeking. These alternatives
>> can pre-initialize note data in that case. This is a very rarely used
>> feature, but avoids the problem that instruments are missing if you skip
>> one channel.
>>
>> Example:
>> We have a track playing a new instrument at row 60. If the user skips
>> this track before arriving row 60, it will not be played at all, thus
>> causing silence. Now consider the normal next row, would change data
>> here or even has a complete empty row (but expecting to play a long
>> looped instrument), in that case it would confuse the actual output.
>> These pointers allow the composer to initialize that in case.
>
> Is this about an human user seeking the file he is playing or a
> tracker formats that specify seeking commands?

The composer can specifiy alternative seek paths if he wants...of
course, if there are formats which also support this feature, it can be
done here as well.

>
>>>>      /** Tempo change or zero to skip tempo change.  */
>>>>      uint16_t tempo;
>>>
>>> Cryptic comment. This is the timestamp where the tempo change? An
>>> index to a list a tempo changing structures? The instrument number
>>> that has a different tempo?
>>
>> Fixed.
>>
>>>
>>>>      /** Played nesting level (GoSub command maximum nesting
>>>> depth).  */
>>>>      uint16_t played;
>>>
>>> Again, BSS?
>>
>> Would require duplicating each track information in the player and
>> therefore add a lot of unnecessary complexity.
>
> So that means that the player _writes_ to this struct? This seems not
> optimal design in a modularity point of view. IMHO, everything that is
> feeded to the player should be read-only and the player should only
> write in his own context structure.

This would duplicate the amount of work required, i.e. duplicating the
whole order list structure. Do you think, this is worth the additional
overhead?

-- 

Best regards,
                   :-) Basty/CDGS (-:

-------------- next part --------------
A non-text attachment was scrubbed...
Name: order.h_20100715.patch
Type: text/x-diff
Size: 7631 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-soc/attachments/20100715/bf594d37/attachment.patch>