Re: [Boost-testing] [dramatec] boost tree corrupted?

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2005-12-17 23:05:09

Jim Douglas writes:
> I am using GNU tar version 1.13 as supplied on the QNX distribution. And 
> likewise, it does not complain about the file - just makes a mess of the 
> names by appending a few extra 'p's.

It unpacks correctly for me on Windows using Cygwin's (GNU) tar
1.13.25, bzip2 1.0.2 and the following command line:

   tar -xjf boost-CVS-HEAD-2005-12-16-17-43.tar.bz2

Just in case, the archive's size before unpacking should be 11,304,911
bytes. 

Googling a bit on the topic reveals this, though (from
http://www.amanda.org/patches/):

    * GNU tar 1.13: even though the second problem in release 1.12 is
      already fixed in 1.13, the other is only partially fixed.
      Besides, there are a couple of other problems in release 1.13,
      related with exclude patterns and restoring, so its use is not
      recommended. Hopefully, all of these problems will be fixed in
      GNU tar 1.14. If you want to use something past 1.12, 1.13.25
      appears to be stable.

So it looks like you might have either to upgrade or downgrade your
tar executable. Alternatively, you could direct regression.py to get
the sources using anonymous CVS access.

HTH,
-- 
Aleksey Gurtovoy
MetaCommunications Engineering