Re: [lzip-bug] FYI: Serious Design Flaws

[Antonio Diaz Diaz]

Dear all,

Two weeks ago, John L. Males reported "serious design issues that cause 
extremely serious consequences", but he refused to explain what issues he 
had found. Later, in a private message he defined the issues as lzip 
compressing much slower than gzip (literally "as in minutes vs days") under 
certain circumstances that he didn't reveal.

As Mr. Males stated that he has no desire to provide any further information 
(not even the options that he passed to gzip and lzip), I'm asking here if 
anybody has ever observed a behavior as the one described by Mr. Males (lzip 
compressing about 1000 times slower than gzip).

If you find such "slow-compressing data", please, tell me so that I can at 
least document how to speed up compression is that case.

As an example of what I'm *not* looking for, I once measured 'lzip -9' 
compressing 56 times slower than 'gzip -9'. It happened with a system log 
file, a kind of file that tends to be very compressible and slow to 
compress. This is normal. 'lzip -9' uses a dictionary size 1024 times larger 
than 'gzip -9'. Therefore it is expected to compress better and be slower. 
In this case the "problem" can be "fixed" by using 'lzip -0', which can 
compress these kind of files better and in about the same time as 'gzip -6'.

-rw-r--r-- 1 19368375 2009-02-12 02:06 warn
-rw-r--r-- 1   966530 2009-02-12 02:06 warn.gz
-rw-r--r-- 1   660487 2009-02-12 02:06 warn-0.lz
-rw-r--r-- 1   328203 2009-02-12 02:06 warn-9.lz


Thanks,
Antonio.