Re: [Bug-tar] new/old extract problems with 6GB sparse file

[Joerg Schilling]

Paul Eggert <address@hidden> wrote:

> Joerg Schilling <address@hidden> writes:

> > As star is able to print the right size, this is mostunlikely for our case.
>
> The file's size is stored elsewhere, so even if star prints the right
> size that doesn't mean that no data went missing.  GNU tar prints the
> right size here too, not that that helps recover data that's not there.
>
> (At least GNU tar doesn't dump core here, like star does. :-)

Do you know what GNU tar would do in case it would be able to extract as 
many data (~ 4 GB) from the archive as star does and did not stop at ~ 1.8 GB 
already?


When storing sparse files, GNU tar violates general structuring rules for TAR
archives, so anything may happen with a broken archive....

One problem with GNU tar sparse archives is that they do not include the correct
"size in archive" value.

Before we know the exact reason for the problem, it does not make sense to 
discuss it.


BTW: the way GNU tar recently started to archive sparse files in POSIX.1-2001
mode violates POSIX.1-2001 (Repeated occurences of the same tag field are 
expected to overwrite former values and a general POSIX.1-2001 rule says that 
vendor unique extensions should be implemented in a way that allows them to 
become part of a future standard...). Even if you ignore these rules, the 
GNU tar method does not allow to archive large sparse files ( > aprox. 1 TB) 
as it quickly overflows the 8 GB limit for the "x" type file).

I thought that you were interested to discuss new extensions before 
implementing them....

For above reasons, I did decide to stay with the old star method for sparse 
files unless I have been able to find a really better solution.


Jörg

-- 
 EMail:address@hidden (home) Jörg Schilling D-13353 Berlin
       address@hidden                (uni)  
       address@hidden     (work) Blog: http://schily.blogspot.com/
 URL:  http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily