I'm inflating compressed HTTP responses from IIS6. IIS6 (by default) supports gzip and deflate, and deflate has less overhead. (Plus most gzip implementations work on the hard disk, and not in memory, which for my purposes would be a travesty...!)
Hey - I spent some time examining the zlib format (actually spent about a week researching compression algorithms) and my understanding is that it CAN be configured to output DEFLATE (RFC 1951), not zlib (RFC 1950). From memory, something about passing a parameter of 16 (instead of 1-15) at a particular point. Anyone know about that? I strictly need DEFLATE RFC 1951, not zlib RFC 1950. (haha any idea how many painful hours it took me to nut that out?) Thanks for the comments guys. -Brendan -----Original Message----- From: Robert Roessler [mailto:[EMAIL PROTECTED] Sent: Monday, 19 September 2005 7:35 PM To: [email protected] Subject: Re: Seperating out the DEFLATE algorithm Brendan Hill wrote: > I have some knowledge of C++, and some knowledge of compression and > encryption algorithms. No expert though. > > I'm wondering what's the best way to take the DEFLATE algorithm > (zdeflate/zinflate) in Crypto++ so it's on it's own. I'd like to use > it in an application, but I don't want to distribute all the other > algorithms with it. > > I tried removing It surgically, but it required various headers, which > required other headers, etc, and eventually I ended up with almost the > entire project again. > > Any tips on how to separate out just one algorithm? Actually, you might consider the code/library available at http://www.zlib.net/ It is free (in multiple senses of the word), and it is what CryptoPP's "z*" compression code is based on. It is also *meant* to be used on its own, so your life will be easier. :) If you have need of higher compression, you could check out the 7-Zip LZMA SDK at http://www.7-zip.org/ Also free, but not as easy to interface with (or maybe it is just the documentation) - but it is possible, and it does work... and gives substantially better compression then the "deflate" algorithm. Robert Roessler [EMAIL PROTECTED] http://www.rftp.com
