[apaddobs]

    my $self = shift;
    my $filename = shift;
    my $fh = new FileHandle ($filename);
    if (not defined ($fh)) {
        $self->message ("$filename: $!");
        return 0;
    }
    my $filesize = (stat ($filename)) [7];
    binmode ($fh);
    $self->_w2 (START);
    my $buf;
    my $written = 0;
    while ($filesize > 0) {
        my $bufsiz = $fh->read ($buf, 512);
        $self->_write ($buf, $bufsiz);
        $filesize -= $bufsiz;
    }
    $self->_w2 (EOF);
    $fh->close;
}

sub _w2 {
    my $self = shift;
    my $val = shift;
    $self->_write (pack ("n", $val), 2);
}

sub _write {
    my $self = shift;
    my $bytes = shift;
    my $nbytes = shift;
    $sock->syswrite ($bytes, $nbytes);
}

Sending a file of 2G takes quite sometime to go through. If I want to compress it over the socket and send, how do i do it? i.e., i don't want to gzip,tar and send the tar.gz over the socket. Coz that will put the overhead of untarring(zipping) at the other end.

--Apaddobs.

[GunnerInc]

Well any type of "compression" that you do on your data even as simple a s RLE, the client still has to undo your compression. Overhead in todays computers is a mute point, either have them sit there and wait for 2gb to download or compress it and let their computer uncompress it... most will vote for the later.

But the question is also, what is in the 2gb of data? Not all data can be compressed as well as your hoping.

[turboscrew]

Isn't there a Zlib or something that lets you compress blockwise?

This post has been edited by turboscrew: 21 May 2012 - 09:08 PM

[JackOfAllTrades]

Moved to Perl

[dsherohman]

There are several compression-related modules available for Perl, it's just a matter of choosing one that uses a compression format that the receiving software will know how to decompress. As GunnerInc said, any compression you do will require the receiving computer to decompress the data, but the computational overhead of doing so is negligible these days.

[apaddobs]

dsherohman, on 23 May 2012 - 05:27 AM, said:

There are several compression-related modules available for Perl, it's just a matter of choosing one that uses a compression format that the receiving software will know how to decompress. As GunnerInc said, any compression you do will require the receiving computer to decompress the data, but the computational overhead of doing so is negligible these days.

Ok so what I am saying is this. If I have a.jar, b.jar and c.jar (each being say close to 1G each), would creating an total.tar.gz (consisting a.jar, b.jar and c.jar) and sending the total.tar.gz over the socket (like i would send a normal file) make sense at all?
I'm looking for something like tar -zxvf a.tar | more (wherein the |more starts doesn't wait for the entire tarring to happen and I can have the "more" part ready to be sent over the socket). Did i explain it correctly?

[GWatt]

You can pipe to ssh and have that write data over the network.

tar cz folder/ | ssh host cat \> /file/on/remote/computer

[apaddobs]

GWatt, on 24 July 2012 - 10:21 AM, said:

You can pipe to ssh and have that write data over the network.

tar cz folder/ | ssh host cat \> /file/on/remote/computer

Correct. The question was how to achieve this in my Perl code wherein I open a socket. That is how to incorporate this in the $fh->read and $socket->write.