Re: [Boost-users] [iostreams] WinRar cannot read zlib archive

From: Jonathan Turkanis (technews_at_[hidden])
Date: 2005-11-16 12:18:28

Christian Henning wrote:
> When I use this little test program, taken from the zlib examples.
> WinRar doesn't have any problem opening it. The generated file
> "chh.gz" is 26 bytes.
>
> int main()
> {
>    gzFile file;
>
>    file = gzopen( "chh.gz", "wb");

Okay, you're writing data in the gzip format, which has a different headder and
footer than the zlib format, even though the body of the compressed data is the
same. For this, you should be using the gzip filters:
www.boost.org/libs/iostreams/doc/index.html?path=4.2.2.11.

>    gzputc(file, 'h');
>    gzputs(file, "ello" );
>
>    gzseek(file, 1L, SEEK_CUR); /* add one zero byte */
>
>    gzclose(file);
>
>    return 0;
> }
>
> Using the iostreams I think, and I might be wrong, should generate the
> same file using this test program:
>
> int main()
> {
>    io::filtering_ostream out;
>    out.push( io::zlib_compressor() );

try:

     out.push( io::gzip_compressor() );

>    out.push( io::file_sink( "chh.gz", std::ios::binary ));
>
>    out << "hello";
>
>    return 0;
> }
>
> Here the file is 13 bytes. So I guess the header or footer is wrong.
>
> Greets,
> Christian

-- 
Jonathan Turkanis
www.kangaroologic.com