[vorbis-dev] question about encoding of lacing values

[Monty]

> I'm reading the logical bitstream framing doc, and I have a question about
> lacing values, which encode the length of a packet (I think).  For example,
> the values
> 
>     255, 255, 243
> 
> encode a length of 753.

correct.

> I'm wondering whether a format more like this was considered
> 
> 	length		bytes		encoding
> 	0-254		1		value in one byte
> 	255-65534	3		one 255 + value in two bytes
> 	65535-16777215	5		two 255 + value in three bytes
> 	etc
> 
> It does cost an extra byte for lengths 255-509, but is smaller for lengths
> over 764.  To me, it seems simpler.

...if you end up with nearly all of your packets in the 255-509 range (as will
be common in vorbis), you lose overall.  The system, as I constructed it, has a
lower fixed worst-case encoding overhead (roughly 1% for packets over 100
bytes), regardless of packet size, and, of course, lowest possible byte
overhead for packets < 255 bytes.  I use even byte values so that the header
interposes an even number of bytes without padding (wasted bits) or repacking
the packets.

Monty

The idea was to minimize 
> 
> --Mike
> 
> -- 
> [O]ne of the features of the Internet [...] is that small groups of people can
> greatly disturb large organizations.  --Charles C. Mann
> 
> --- >8 ----
> List archives:  http://www.xiph.org/archives/
> Ogg project homepage: http://www.xiph.org/ogg/
> 

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/