linux fan wrote:
> On 3/19/10, Mike McCarty <mike.mcca...@sbcglobal.net> wrote:
> 
>> I wonder if "head" is closing the input pipe when it has read
>> all it needs, and that's causing the error. I can't reproduce
>> that problem with my host system, however.
> 
> It is tar-1.23 and not head
> 
> head --version
> head (GNU coreutils) 8.4
> 
> tar -xf tar-1.23.tar.bz2
> cd tar-1.23
> ./configure --prefix=/usr --bindir=/bin --libexecdir=/usr/sbin
> make
> 
> src/tar --version
> tar (GNU tar) 1.23
> src/tar -tf ../tar-1.23.tar.bz2 | head -1
> tar-1.23/
> src/tar: write error
> 
> tar --version
> tar (GNU tar) 1.22
> tar -tf ../tar-1.23.tar.bz2 | head -1
> tar: Record size = 8 blocks
> tar-1.23/

Right.  They changed something with regard to the SIGPIPE signal.  The 
default action is to terminate (man 7 signal), but they changed it to 
SIGIGN (ignore).  It needs to be fixed.

http://git.savannah.gnu.org/gitweb/?p=tar.git;a=commitdiff;h=release_1_22-41-g0c6c288

If I revert this change, the program works normally.  I did not revert 
the change to the tests and test 91 failed.

I did report the error, and it seems it was also reported earlier today 
by someone else.  Right now it is only an irritation, but we may want to 
consider reverting to version 1.22.

   -- Bruce
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page

Reply via email to