[Zlib-devel] Plans for zlib 1.2.5

[Török Edwin]

On 02/22/2010 06:05 AM, Mark Adler wrote:
> zlibbers,
> 
> As 1.2.4 is (hopefully) close to going out the door, my plans for 1.2.5 (or maybe it will be called 1.3.0) are:
> 
> 1.  Add rsyncable option to periodically flush as a function of the input data for faster updating of large compressed files with rsync.
> 
> 2.  Add parallel processor support for compression.
> 
> 3.  Add optional compressed stream indexing to make gzseek() useful in the reverse direction when reading.
> 
> Any other suggestions?

I noticed that Z_FILTERED gives a significant improvement in compression
ratio (around 9%) when used on text files which contain lot of
hexadecimal-encoded data; for example ClamAV's database, which uses
Z_FILTERED now for compression.

Could zlib automatically determine what is best for a block, and use the
strategy that gives best compression ratio? Is there anything else that
could be tuned by introducing new strategies that are transparent to the
decompressor?

Best regards,
--Edwin