pax upgrade to 1003.2/D10

[fredriks at ihlpm.att.com]

Submitted-by: fredriks at ihlpm.att.com

I am trying to upgrade Mark Colburns PAX utility to be 1003.2/D10
conformant(and to run under MINIX). In working with it I have run 
accross an inconsistancy (at least an ambiguity) in the PAX 
description.

1003.2 says that "The archives created by PAX can be concatinated to
combine multiple volumes on a single tape or file."

Thats simple enough. 1003.1 says that "At the end of the archive file
are two blocks filled with binary zeroes, interpreted as an end-of-
archive indicator"

Well, if I concatinate two archive files using cat(1), I have two
end-of-archive markers in my archive! That causes problems. Right now
PAX(Mark Colburns) quits after it finds a EOA marker, 
so that it validates quote 1.
Also, PAX right now doesn't append right. It basically does what
quote 1 says, conseptually concatinating two archives, but since it
quits once it finds the EOA marker, you can never access the appended
stuff.

One solution is to ignore the EOA marker and read until you get a 
 physical EOF. If that is done, how are you supposed to differentiate
between multiple volumes. The only solution I see is that you assume
that multiple volumes do NOT have a EOA marker before EOF of volume X.

As you can see, having multiple EOA in the archive file, is a MESS!

Lars
L.Fredriksen at att.com

PS! This has absolutely NOTHING to do with AT&T

Volume-Number: Volume 21, Number 130