Section packing

I've just started to read the list and would add some more details about the section packing issue. The given response does not reflect the issue very well.

The main advantage of section packing it not eliminating the four header bytes, but to replace the padding bytes (0xff) of a TS frame (if not fully filled with a section) with useful section data. I am attaching a sample comparison between a packed and a non-packed stream (which I copied for our encapsulator manual). Keep in mind, that the pointer byte is not part of a section, but is placed just behind the TS header (PUSI will be set in this frame), if a section starts in a TS frame (independent where in the TS frame). The pointer byte just tells the upper layer, where to find the start of the section.

If you have a well utilized DVB channel, you will not recognize a big difference between a packed (delayed) or un-packed (wasting capacity) channel, because as soon as a new section is created it will be packed into the waiting TS frame. But: The more different PIDs you use, the less efficient are the packed channels (timeouts may occur (last frame will be send with padding bytes), because no new section arrives on this PID channel).

Karsten

data planet international AG