I wanted to find the difference between accessed and modified epochs of a bunch of files. The incantation below gives you this number for each item in a directory:

$ stat -c "%X %Y %N" ~/* | awk '{diff = $1 - $2; print diff, $3}'
40449 '/home/home/my_cgi'
2173034 '/home/home/run_pvr.sh'
124618 '/home/home/tools'

powerpc-eabi-ranlib: command not found

I'm trying to compile wiimc, these two pages have proved very helpful setting up my compile environment and getting to the point of building wiimc. However, my make in wiimc directory falls over complaining:

libtool: install: chmod 644 /opt/devkitpro/portlibs/ppc/lib/libfribidi.a
libtool: install: powerpc-eabi-ranlib /opt/devkitpro/portlibs/ppc/lib/libfribidi.a
../libtool: line 947: powerpc-eabi-ranlib: command not found

The solution I found is to run make with dir1, dir2 and dir3, then to cd into the directories for dir1, dir2, dir3 and run make install separately. Like this:

$ make dir1; make dir2; make dir3
$ cd ./fribidi; make install; cd ..
$ cd ./libexif; make install; cd ..
$ cd ./libiconv; make install; cd ..

This is suitably hacky for my needs.

Compiling No Hands on F17

I'm back on F17, on at least one machine I use regularly. I've been looking at No Hands that will allow me to use my desktop as a hands-free set for my mobile. The mobile communicates with the desktop using bluetooth.

I've got it to compile by installing dependencies piece-meal and using the following incantation:

LDFLAGS="-lpthread `pkg-config -cflags --libs dbus-1`" ./configure CXXFLAGS=-fpermissive

I haven't actually got the code working yet (seems I should fiddle with bluetooth?) Any suggestions are welcome.

connecting a tunnel to a reverse ssh connection and mounting with sshfs

I have a machine (treestump) that connects to a server with a known ip address (applecart). treestump opens a reverse proxy connection onto applecart at port 29999. From applecart, I can connect to treestump from applecart by issuing: 'ssh me@localhost -p 29999'. I have password less logins configured.

I'm on my machine here (rainplant). rainplant can't see treestump but does have an account and passwordless login to applecart. I want to mount treestump using sshfs. This requires two steps.

1. Set up a tunnel all the way through to treestump.

    ssh -L localhost:1999:localhost:29999 me@applecart

1.a Test the connection.

    ssh me@localhost -p 1999

2 Mount using sshfs

    sshfs -p 1999 me@localhost:/storage/ ./treestumpstorage

I saw a suggestion to use autossh, this makes sense to me on treestump.

For what it's worth: I've been really enjoying gnome3 and I like the direction it's taking.

yui-compressor from rhino

I'm using rhino for a build dependency on a project I'm working on. Rhino is building a javascript application for me and part of that process is running yui-compressor - so this post is about calling yui-compressor from Rhino.

I couldn't get loadClass to accept the 'resolve' boolean so I've been reduced to the following to get something working:



 var COMPRESS = new Object();
    COMPRESS['yuicompress'] = ["-jar", JSLIBS+'/bin/yuicompressor-2.4.8pre.jar'];
var YUICOMPRESS_OPTIONS = [
  "--nomunge", //               Minify only, do not obfuscate
  "--preserve-semi", //          Preserve all semicolons
  "--disable-optimizations", //   Disable all micro optimizations
];
     var compressCMD = COMPRESS['yuicompress'];
    if (YUICOMPRESS_OPTIONS.length > 0) {
        if (YUICOMPRESS_OPTIONS.length == 1) {
            compressCMD.push(YUICOMPRESS_OPTIONS);
        }
        else {
            compressCMD.concat(YUICOMPRESS_OPTIONS);
        }
    }
    compressCMD.push([infile]);
    var options = new Array();
    options['args'] = compressCMD;
    options['output'] = "";
    runCommand("java", options);

    //print(options['output']);
    var w = new java.io.FileWriter(outfile);
    try
    {
        w.write(options['output']);
    }
    finally
    {
        w.close();
    }
    return;

routing real-time audio to the internet

I've been working on a server-side audio tool. My app works well on the desktop - it is written in Java and the audio from SourceDataLine just works. So, I moved the app to a remove server.

Pulseaudio was installed on the remote server, and I thought it should be pretty easy to route the output back over the internet to myself.

In the end it was:

parec -r | oggenc -r -o - - | ./sighttpd -f ./oggout.conf

I called my java code with 'padsp', I'm not sure that is necessary but it probably is. This method avoids using Gstreamer (that should be a good solution if you want to scale: It can plug your audio into icecast/shoutcast.)

SourceDataLine provided the audio stream in my Java app.

NOTE: I spent a whole bunch of time playing with a file output pipe from pulse audio. The excellent support from pulse audio mailing list and irc channel steered me away from that.