Re: [Lzip-bug] Packaging lzip

[Antonio Diaz Diaz]

Hello Bram,

Bram Neijt wrote:
> After reading about lzip in this months Linux Format and seeing that it
> was not yet packaged, I decided to try packaging it for the Ubuntu
> Universe repository.

Thank you very much.


> With lzma already packaged and other compression algorithms already in
> the packaging queue (lrzip for example) the actual package may not make
> it into the Ubuntu Universe repository.

If the lzma already packaged is one that produces lzma-alone files it 
would be better to replace it with lzip. See this quote from 
http://lists.gnu.org/archive/html/lzip-bug/2008-11/msg00003.html :

"Lzip provides a much simpler and reliable implementation. Lzip also 
provides a simple but safe file format, with magic bytes and integrity 
checking. The stable branch of lzma-utils, the one "widely used", uses 
the lzma-alone file format, which lacks both. I hope everybody who cares 
about data safety will switch from the lzma-alone format to the lzip 
format as soon as they know about it."


> In light of this LZMA already being in the repositories, it might be
> good to create a table which compares lzip to the other compression
> algorithms. lrzip for instance shows off better compression then LZMA
> but it doesn't mention it's speed.

Lrzip does have significative differences from gzip/bzip2/lzip:
- It is optimised for large files, especially larger than 100MB.
- It only works on single files.
- Is not really usable (for compression) with less than 256MB (of RAM).
- Does not work on stdin/stdout.
- Lacks integrity checking (and it will be quite hard to implement).


> A table containing the compression ratio, compression speed and
> decompression speed might be good to post on the lzip website.

There are specialized benchmark sites like 
http://www.cs.fit.edu/~mmahoney/compression/text.html or 
http://www.maximumcompression.com/index.html with much more and better 
data that what I can test myself.


Best regards,
Antonio.