Buffers are the basic unit of data transfer in GStreamer. The Buffer type provides all the state necessary to define a region of memory as part of a stream. Sub-buffers are also supported, allowing a smaller region of a buffer to become its own buffer, with mechanisms in place to ensure that neither memory space goes away prematurely.
Non-plugins will usually not need to allocate buffers, but they can be allocated using new {@link #Buffer(int)} to create a buffer with preallocated data of a given size.
The data pointed to by the buffer can be accessed with the {@link #getByteBuffer}method. For buffers of size 0, the data pointer is undefined (usually NULL) and should never be used.
If an element knows what pad you will push the buffer out on, it should use gst_pad_alloc_buffer() instead to create a buffer. This allows downstream elements to provide special buffers to write in, like hardware buffers.
A buffer has a pointer to a {@link Caps} describing the media type of the datain the buffer. Attach caps to the buffer with {@link #setCaps}; this is typically done before pushing out a buffer using gst_pad_push() so that the downstream element knows the type of the buffer.
A buffer will usually have a timestamp, and a duration, but neither of these are guaranteed (they may be set to -1). Whenever a meaningful value can be given for these, they should be set. The timestamp and duration are measured in nanoseconds (they are long values).
A buffer can also have one or both of a start and an end offset. These are media-type specific. For video buffers, the start offset will generally be the frame number. For audio buffers, it will be the number of samples produced so far. For compressed data, it could be the byte offset in a source or destination file. Likewise, the end offset will be the offset of the end of the buffer. These can only be meaningfully interpreted if you know the media type of the buffer (the #GstCaps set on it). Either or both can be set to -1.
To efficiently create a smaller buffer out of an existing one, you can use {@link #createSubBuffer}.
If a plug-in wants to modify the buffer data in-place, it should first obtain a buffer that is safe to modify by using {@link #makeWritable}. This function is optimized so that a copy will only be made when it is necessary.
A plugin that only wishes to modify the metadata of a buffer, such as the offset, timestamp or caps, should use gst_buffer_make_metadata_writable(), which will create a subbuffer of the original buffer to ensure the caller has sole ownership, and not copy the buffer data.
Buffers can be efficiently merged into a larger buffer with gst_buffer_merge() and gst_buffer_span() if the gst_buffer_is_span_fast() function returns TRUE.
|
|
|
|
|
|
|
|
|
|
|
|