On Thu, May 18, 2006 at 11:05:38PM +0200, Florent Daigni?re (NextGen$) wrote:
> * Ian Clarke <ian at locut.us> [2006-05-18 13:11:34]:
> 
> > High level question:  Why is it necessary to support several  
> > compression standards?
> 
> I'm not sure either that it is a good idea.

It is vital to support multiple compression standards for
*decompression*. One more reason: In future we may add a new compression
algorithm! Whether it is vital to try every possible algorithm for
compression is an open question.
> 
> NextGen$
> 
> > Wouldn't it be simpler just to support one  
> > (ie. the one that achieves the best compression, without suffering  
> > from legal issues)?
> > 
> > Ian.
> > 
> > On 18 May 2006, at 11:25, Matthew Toseland wrote:
> > 
> > > http://www.7-zip.org/sdk.html - there is a java version.
> > >
> > > We should provide this as a compression codec.
> > >
> > > Thus, we can provide gzip, bzip2 and 7zip compressors, plus zip and  
> > > tar
> > > archivers.
> > >
> > > I am currently testing some code which will allow larger freesites  
> > > to be
> > > inserted, the next job after that is ZIP manifest support, which  
> > > judging
> > > from the test site should make a substantial difference.
> > >
> > > I am currently inserting www.hardwarebook.net, just the HTML. On disk
> > > this is 3.2MB. Inserting it at present uses 489 blocks = 16MB. As a  
> > > zip
> > > file it would be 24 blocks plus 12 check blocks plus 1 redirect. As a
> > > tar.gz it would be 9 blocks plus 5 check blocks plus 1 redirect. As a
> > > tar.bz2 or a tar.7z it would be 6 blocks plus 3 check blocks plus 1
> > > redirect.
> > >
> > > Raw       489
> > > tar       133
> > > zip       37
> > > tar.gz    15
> > > tbz/t7z   10
> > >
> > > Admittedly this wouldn't be quite so extreme if we included the  
> > > images.
> > >
> > > Obviously the big gain is from raw to zip. This is already implemented
> > > for fetching but not yet for inserting. There were some legal issues
> > > with the tar/bz2 libraries but these have been sorted out thanks to
> > > GPL3's explicit compatibility with the ASL2.
> > >
> > > We probably don't want the whole site in one big archive. One
> > > possibility would be to split it by content type: one container  
> > > includes
> > > all the HTML, one includes all the GIFs etc, and files over a certain
> > > size are inserted separately.
> > >
> > > One other minor point of note: We need to provide a means for  
> > > clients to
> > > be able to tell us not only whether to not try to compress at all, but
> > > also whether to try the really heavy compressors.
> > > --
> > > Matthew J Toseland - toad at amphibian.dyndns.org
> > > Freenet Project Official Codemonkey - http://freenetproject.org/
> > > ICTHUS - Nothing is impossible. Our Boss says so.
> > > _______________________________________________
> > > Devl mailing list
> > > Devl at freenetproject.org
> > > http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
> > 
> > _______________________________________________
> > Devl mailing list
> > Devl at freenetproject.org
> > http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
> > 



> _______________________________________________
> Devl mailing list
> Devl at freenetproject.org
> http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

-- 
Matthew J Toseland - toad at amphibian.dyndns.org
Freenet Project Official Codemonkey - http://freenetproject.org/
ICTHUS - Nothing is impossible. Our Boss says so.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20060519/e1c07847/attachment.pgp>

Reply via email to