Re: [E-devel] fast decompression lib

[Carsten H. (T. R. <ra...@ra...>]

On Tue, 18 Aug 2009 07:11:43 +0200 (CEST) Vincent Torri <vt...@un...>
said:

> >>>> while looking at the doc of UPX, i've found that lossless
> >>>> (de)compression library:
> >>>>
> >>>> http://www.oberhumer.com/opensource/lzo/
> >>>>
> >>>> which says to compress better and to be faster decompression than
> >>>> zlib.
> >>>>
> >>>> There is no comparison numbers, but maybe it is worth trying to
> >>>> compare zlib and that lib to see if it would be better to use it in
> >>>> eet.
> >>>
> >>> There are other contenders, generally more modern compression schemes
> >>> that are generally ignored coz things like zip, gz, and bz2 are
> >>> too firmly embedded in our conciousness.  Too much historical inertia
> >>> is not good.
> >>>
> >>> So, if this is to be done, a more thorough roundup of them all would be
> >>> the way to go.
> >>
> >> Maybe it's not so good for us: that library is GPL. If I'm not mistaken,
> >> that would mean that eet must be GPL too, right ?
> >>
> >> Vincent
> >>
> >
> > Yes, this is under a GPLv2+ license with additional commercial
> > licensing, so I don't see any chance to use it in any non-GPL or
> > non-commercial application - except calling the binary directly
> > without including any sourcecode itself.
> >
> > But there are alternatives: The best one is most likely XZ (the
> > successor of lzma) which usually compresses at a 2-digit percentage
> > better compared to gzip at the cost of slightly more computing time to
> > decompress and a lot more to compress (still better than bzip2 in any
> > case).
> > It's available under LGPL so, there should be no problems with the license.
> > http://tukaani.org/xz/
> 
> gzip also consumes few memory when decompressing, which is good for 
> embedded devices. So comparison must be made on speed and memory 
> consumption.

also zlib is already on and available for many many many systems... :)

-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    ra...@ra...