homepage

hi there,
it seems tarfile in python 3.2.2 (as installed in archlinux, but I don't see any additional patch applied on top of the vanilla sources:
http://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/python
) has troubles giving the complete content of a tar ball.
see:
$ wget http://downloads.sourceforge.net/sourceforge/boost/boost_1_44_0.tar.gz
$ md5sum boost_1_44_0.tar.gz 
085fce4ff2089375105d72475d730e15 boost_1_44_0.tar.gz
$ python --version
Python 3.2.2
$ python2 --version
Python 2.7.2
$ python ./foo.py
>>> 8145
$ python2 ./foo.py 
>>> 33635
where foo.py is:
##
import tarfile
o = tarfile.open("boost_1_44_0.tar.gz")
print(">>> %s" % len(o.getmembers()))
o.close()
## EOF ##
is it a known bug ?
(this of course prevents TarFile.extractall to be useful w/ python3...)
-s

one interesting additional piece of information is that if I un-tar that file and re-tar it w/o gzip compression, getmembers gets the right answer.
-s

New changeset 341008eab87d by Lars Gustäbel in branch '3.2':
Issue #13158: Fix decoding and encoding of base-256 number fields in tarfile.
http://hg.python.org/cpython/rev/341008eab87d
New changeset 158430b2b552 by Lars Gustäbel in branch 'default':
Merge with 3.2: Issue #13158: Fix decoding and encoding of base-256 number fields in tarfile.
http://hg.python.org/cpython/rev/158430b2b552 

Thanks for the report. There was a problem decoding a special and rare kind of header field in the archive. The format of the archive is of very bad quality BTW ;-)

thanks!
> The format of the archive is of very bad quality BTW ;-)
well, that's C++ :P
-s