🚀 go-pugleaf

Matt DiMeo (mdimeo@brooktree.com) wrote:
: In article <3hvs5p$abq@news.bu.edu>, Nik Conwell <nik@bu.edu> wrote:
: >The gzip TODO file indicates that one of the planned features is a library of
: >routines that can be called to do compressed I/O.  Does anybody know of the
: >status of this, or if there are routines available?

: Dunno.  But I've written a relatively simple gzopen() function that forks
: off a gzip -d process and reads from a pipe.  I've found that in some heavily
: disk-io-bound programs, it's actually faster to read (300+ meg) files in
: gzipped form, even with the piping and decompressing.

: If anyone wants the code, send me mail.

Unfortunately, my application requires unzipping *many* gzipped files, so
doing many fork/popen() calls seems wasteful.

There's also a sample program in the gzip 1.2.4 sample directory which does a
popen() to a gzip process...

% pwd
/usr/local/src/gzip-1.2.4/sample
% head zread.c 
#include <stdio.h>

/* Trivial example of reading a gzip'ed file or gzip'ed standard input
 * using stdio functions fread(), getc(), etc... fseek() is not supported.
 * Modify according to your needs. You can easily construct the symmetric
 * zwrite program.
 *
 * Usage: zread [file[.gz]]
 * This programs assumes that gzip is somewhere in your path.
 */

--
Nik Conwell                 Boston University                   nik@bu.edu