Post a Comment On: cbloom rants

Oodle now ships with "ozip", a command line compressor that acts like gzip (and bzip, and xz). ozip can be used as a command line compressor to create or decode Oodle-compressed files; ozip can also be used to pipe Oodle-compressed streams. The intention is that ozip should act similarly to gzip (in terms of command line arguments, but with more compression and faster decompression) so it can be dropped in to standard work flows that use gzip-like compressors. For example ozip can be used with tar "I" ("--use-compress-program") to pipe the tar package through ozip. ozip works with pipes (particularly useful on Unix), so it can be used to pipe compressed data. (eg. with things like "zfs send" to a pipe). A pre-compiled ozip binary is now distributed with the Oodle SDK. You can also get and modify the ozip source code on github : ozip on github To build ozip from source code, you need the Oodle SDK . (ozip is open source and public domain, Oodle is not). If you have corrections to ozip, we're happy to take pull requests, particularly wrst making sure we act like gzip and it's easy to drop in ozip to gzip-like work flows on Unix. When writing ozip, we found that gzip/bzip/xz don't have the exact same command line argument handling, and yet are treated as interchangeable by various programs. We tried to replicate the common intersection of their behavior. ozip is a single file compressor, not a package archiver. It does not store file names or metadata. ozip was written by my brother, James Bloom. If you have server workflows that involve streaming compressed data, or think you could benefit from Oodle, we'd be happy to hear from you. We're still evolving our solutions in this space. If you are compressing very large files (not piped streams), you can get much higher performance by using threads and asynchronous IO to overlap IO with compression CPU time. If this is important to you, ask about the "oozi" example in Oodle for reference on how to do that.