Re: [Bug-tar] [PATCH] p7zip support

[Robert Millan]

On Wed, Jan 11, 2006 at 11:15:03AM -0600, Doug McLaren wrote:
> On Mon, Jan 02, 2006 at 06:40:39PM +0100, Robert Millan wrote:
> 
> | > Sure, that's temporary space. But anyway, one must have twice as much
> | > space available when restoring from a 7za archive than when restoring
> | > from an archive compressed with another compression method. And that's
> | > a big drawback.
> 
> It's also just plain ... nasty.

Yes, but it's isolated behind the p7zip(1) wrapper.  I don't think tar should be
concerned about implementation details of the compressor as long as it provides
a sane command-line interface to interact with (which p7zip(1) does).

> |   - It depends on what you do.  In my example (thousands of small files), 
> you
> |   get a significant gain in permanent space for a minimal cost in temporary
> |   space.
> 
> Wouldn't `tar -zcvf' or `tar -jcvf' benefit similarly (but just using
> gzip or bzip2 rather than p7zip?)

I was explaining why the lack of streaming support is not always a problem.
This question isn't applicable to gzip or bzip2 since they do have proper
streaming.

As for the comparison (with thousands of small files), see the link I pasted
before:

  http://www.linuks.mine.nu/sizematters/
 
> Perhaps, but it just doesn't seem ready for tar use today.

Ok.  If my arguments didn't convince you, would you consider adding the part of
my patch that would allow to auto-detect .tar.7z archives ? (patch attached)

> And as was discussed in the previous thread, you can use your p7zip
> wrapper with tar today with no patches -- just use 
> 
>    tar --use-compress-program=p7zip ...

This flag is useful (although not ideal).  I'll write it down.

-- 
Robert Millan

p7zip.diff
Description: Text document