[standards-jig] JNG - binary headers.

Timothy Carpenter timbeau_hk at yahoo.co.uk
Wed Aug 14 18:39:19 UTC 2002


I too have been lurking, but want to voice my USD0.02

Binary headers need not be fixed or impossible to evolve 'elegantly'.

Headers can be extensible using bits, or even bytes to flag an extension
much as UTF-8, RCSU and UNICODE work together and at different levels to
allow compact bytestreams for ASCII whilst allowing for the full character
sets and text-friendliness with manageable overheads and consistent encoding
after unstuffing.

Trailing binary sentinels (i.e. a post-message stamp to flag EoFrame) are
bound to occur randomly in the data stream and would cause the computer
equivalent of dysentry. I also do not think it sensible for each reader
process to have to scan and compare every byte to detect the end of what
they are getting before having to process it.

Thus I support the proposition for dual standards so that link level
protocols can be selected (e.g. binary headers on or off) via negotiation or
alternate ports.

<inspiration>
They managed to encode colour PAL TV into the existing B/W signal back in
1965 so old B/W and new Colour TVs could use the same transmission (and it
is better than Never Twice the Same Colour). They have since encoded extra
lines into the 'black bars' around widescreen movies to allow both 4:3 sets
and 16:9 sets to continue to tune to the same signal and for the 16:9 TVs to
display full 625 lines (605 nominal) resolution while old, unaware 4:3 TVs
just show those lines as the black bars in the 'letterbox' format. it is
called PALplus, I believe.
</inspiration>

Everything is possible if we are smart enough.
Tim

__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com




More information about the Standards mailing list