Re: [Duplicity-talk] Re: Help me replace tar!

[Ben Escoto]

On Tue, 21 Oct 2003 14:31:10 +0200 (CEST)
Joerg Schilling <address@hidden> wrote:
> OK, I started to document the rest of the star format. It is more than
> twice the amount it was yeaterday but it is still far from being complete...
> 
>       http://www.fokus.fraunhofer.de/usr/schilling/private/man/star.4.html
> 
> BTW: I am not sure if ot makes sense to write a tar implementation
> in an interpreter language as you seem to do. The speed of such a
> beast nowerdays mainly depend on the efficience of handling archive
> headers. Star is higly optimized for doing so even from the
> perspective of a C program.

Well, I never wrote a tar implementation.  Duplicity is mainly a
script for organizing and encrypting tar-style volumes, and then
sending and retrieving them over a network.  Most of the code that
does the "heavy lifting" is farmed out to other packages like GnuPG
and librsync that are written in C.

I didn't write the tar processing either, I got this from Lars
Gustabel, who wrote a python tar library, I think mainly from the GNU
docs I mentioned.  I (at first) found it convenient to just
incorporate his library into my program.  There was no python cpio or
python star library to use.

Now that you have documented more of the star format it at least makes
it much easier for people like Gustabel to write packages that process
them, like he did with GNU tar.

BTW, I haven't had a chance to read the SUS pax spec yet, but thank
you for referring me to it.


-- 
Ben Escoto

pgpgJdeFY8Zhe.pgp
Description: PGP signature