On May 22, 2010, at 3:58 AM, Benoit Daloze wrote:

> Hi,
>=20
> On 22 May 2010 01:44, Caleb Clausen <vikkous / gmail.com> wrote:
>> I'm not a big fan of .zip either, but give it its due.
>>=20
>> Zip has the advantage that you can 'seek' to arbitrary files within
>> the archive without needing to decompress all the other files that
>> come before it. That's not true of gzipped tar files.....
>>=20
>> Because each member of a zip file is a compressed separately. But =
this
>> same design decision is why .tar.gz compresses better than .zip, =
given
>> the same collection of files.
>>=20
> What about LZMA (p7zip, xz) ?

I have looked into this in the past and could not find any open source =
libraries for reading 7z or xz files. That is the biggest reason not to =
use them, we don't want to start this project off writing a container =
format.

 - Evan

>=20
> =46rom my experience it's far better in size than .zip and =
.tar.{gz,bz2},
> and it has the ability to get one file at a time.
>=20
> The downside, is, I suppose, the speed/computation for =
compressing/extracting.
>=20
> Let's see a little on Ruby's source (I could not get xz working on =
OSX):
> (format, size, compression time, extraction time, relative size to
> original, Mo compressed/s, Mo extracted/s)
> fmt   size   ctime xtime  psize   Mc/s Mx/s
> 7za  5476 14.7s 7.8s 13.01%  2.86  5.39
> bz2  6792  8.7s  3.6s 16.14%  4.84  11.69
> gz    8252  3.0s  4.0s 19.61% 14.03 10.52
>=20
> I personally prefer a little slower compression/extraction with a low
> size, as it make easier downloads.
>=20
> Regards,
> B.D.
>=20
>=20