Re: [Lzip-bug] Want to Jettison xz(1), But Size Matters.

[Ralph Corderoy]

Hi Antonio,

> As you can see, lzip adjusts the dictionary to the file size.

Yep.

> >      $ stat -c '%s  %n' * | sort -k1,1n -k2
> >      20957117  foo.xcf.lzip-m64-s64MiB
> >      21001368  foo.xcf.xz-9
>
> You may perhaps obtain slightly better results with the shorter
> command 'lzip -9s26'.

Right, maximum `-m' of 273, with same `-s 64MiB'.  I settled on `lzip -9
-s 96MiB' in ~/bin/tolz because 96 MiB × 10 is about the maximum I can
dedicate to compression on this particular machine, and I'm happy that
96 MiB, plus a bit, will always be available for decompression.

> >      2^12 to 2^29 bytes. Note that dictionary sizes are quantized.
> >      If the specified size does not match one of the valid sizes, it
> >      will be rounded upwards by adding up to (BYTES / 8) to it.
> >
> > Could info's last sentence be extended slightly with a clue why?
>
> It has nothing to do with lzip's algorithm. It is simply for
> efficiency.  As the dictionary size is just the minimum size of the
> buffer needed to decompress a file, it does not hurt to allocate a
> slightly larger buffer. This allows the size to be coded in just one
> byte, instead of the four bytes used by lzma-alone.

Ah, `DS (coded dictionary size, 1 byte)' in the info.  Perhaps a
reference to that part of the format could be added to the `rounded
upwards' above.

> BTW, do you know zutils' zupdate?
> http://www.nongnu.org/zutils/zutils.html

No, I wasn't aware of zutils at all.  I'm familiar with zgrep(1), etc.,
and here on Arch Linux they're present from the gzip package so zutils
would clash as there's no `alternatives' system like Debian.  Also, I'm
familiar with tolz's logic and tests so I'm happy to stick with that.

BTW, I noticed the info here has a few mispelt words.

    ✗  Additionaly posible substract substracting
    ✓  Additionally possible subtract subtracting

-- 
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy