[Help-tar] problem with GNU tar java tar compatability

[Dan Mayer]

I am trying to add GNU multisession tar support to the java tar package.
I have figured out that my problem comes from

in buffer.c:
if (real_s_totsize - real_s_sizeleft
!= OFF_FROM_HEADER (cursor->oldgnu_header.offset))
{

I added outputs and learned that real_s_sizeleft is not what i was
expecting. I can't figure out where in the header this number is coming
from. I know that the only other file i can find that refrences it is
exctract.c 

this is the last thing i need to make it work. I have all of the proper
data split into the proper filesizes. Any help would be very helpful. 
 
As best I can tell that number comes from how much gnutar actually reads
from the split file from the first disk (cause if I write random data
the number I am of by goes down.) It seems that GNU tar thinks it is
reading slightly more data of the disk than I believe to have written.
It is coming up with a smaller number than what I am expecting to have
to write on the second disk.
 
How is real_s_sizeleft counted? Where is it coming from? How could the
amount of data I am writing actually be wrong? When the file matches up
perfectly as do the file sizes I am calculating? Is it cause I am
closing the first tar with null chars should the data just stop and have
nothing more written after the last char?
Thanks for all your help.

Thanks so much,
Dan