We all know that passwords aren’t very secure. If you choose a password that is easy to remember then its easier to guess via brute force. If you choose a password that is random or hard to remember then you are more likely to write it down. Any well-versed digital community member already has many username/password credentials to remember so we are less likely to remember extremely difficult random passwords. This is the password paradox, which leads me to look into managing access via public/private key pairs instead.

Generate a public/private key pair (if you don’t have one already)

Github has an excellent tutorial with good illustrations on how to generate ssh keys with ssh-keygen, but here is an overview:

• *Backup your existing keys* — if they exists so that you don’t overwrite them.
• generate a new key with ssh-keygen

Definitely setup a passphrase! – Its like a password for your private key. The passphrase is a second line of defence if anyone were to acquire your private key.

The .pub is your public key, you can safely share this anywhere, whereas the private key is entirely private! DO NOT show anyone, don’t copy it anywhere and only securely back it up.. This is your new password and large liability if it fell into the wrong hands.

Example:
Some folks like to generate and manage keypairs for each location, or at least manage certain levels of keys, but I’m not going to dive into that topic.

Configure the Server (if needed)

Make sure that the server has Public key authentication enabled (most do). for OpenSSH it would be the following in the sshd_config:

RSAAuthentication yes
PubkeyAuthentication yes
AuthorizedKeysFile .ssh/authorized_keys

Then all you have to do is:

• Put your public key (the one ending in .pub) as a line in the ~/.ssh/authorized_keys file (create it if it doesn’t exist)
• Restart your SSH server
• Log in without being prompted for a password

This is really handy for managing servers especially on Amazon’s EC2, local access between machines, SFTP, SSH tunneling, or even getting access to a jailbroken iPhone or any other device with OpenSSH installed on it.

The guys over at debian-administrator.org wrote a good guide to Password-less logins with OpenSSH that is worth looking into if you have issues or want to dive deeper.

The Arduino IDE (hosted on google code) doesn’t work with ChipKit so They remade the the IDE and called it MPIDE (Multi-Platform IDE) and its hosted on Github. The released binaries are found here.

They have a decent wiki page on how to run MPIDE which helped.

The article cited issues with 64 bit systems:

64bit Systems

The binary packages for MPIDE are all built for 32bit Linux distributions, if you are running 64bit you need to have some additional libraries installed. You’ll need to have 32bit versions of libelf.so and libreadline.so to be able to program the chipKIT, these need to be in /usr/lib32. You may well need to sym-link these latest versions manually, installing these libs is a fairly advanced operation and will vary significantly between distributions.

So this is what I had to do to get it running on 64 bit Ubuntu (Maverick/10.10):

1) Download the Linux Version and unpack it somwehere.

2) Getting librxtx-java and other jni items loadable via the LD_LIBRARY_PATH or CLASSPATH, which I wrote a post about.

3) The next issue that I ran into was a permissions issue

Cannot run program “/home/username/mpide-0022-chipkit-linux32-20110619/hardware/pic32/compiler/pic32-tools/bin/pic32-g++”: java.io.IOException: error=13, Permission denied

I initially solved it by chmodding everything in that directory, but that resulted in:

pic32-g++: error trying to exec ‘cc1plus’: execvp: No such file or directory

So I followed the advice in this forum post and executed chmod +x  -R ./hardware/pic32/compiler/ and that seemed to do it for me.  I think I was able to compile/verify the blink example at this point.

4) Add 32bit versions of libelf and libreadline6 to /usr/lib32:

The first error I ran into was:

/home/username/mpide-0022-chipkit-linux32-20110619/hardware/pic32/compiler/pic32-tools/bin/../pic32mx/bin/gcc/pic32mx/4.5.1/cc1plus: error while loading shared libraries: libelf.so.1: cannot open shared object file: No such file or directory

I solved this by downloading the i386 deb from here, opening it with archive manager and copying /usr/lib/libelf-0.147.so to /usr/lib32/libelf-0.147.so. after this I creating the versioned symbolic link via ln -s libelf-0.147.so libelf.so.1

Tried it again and was onto my next error:

/home/username/mpide-0022-chipkit-linux32-20110619/hardware/tools/avrdude: error while loading shared libraries: libreadline.so.6: cannot open shared object file: No such file or directory

Repeat the same process for libreadline6 (which contains another required lib too) And I was able to upload the and run the blink example.

Now on to making the TLC Library work..

Synergy2 comes with a GUI on Windows, but for Linux or Apple/Mac OSX there isn’t a GUI packaged with synergy itself. I’ve since found the Quick Synergy GUI to work fairly well between Windows and Linux since the key mappings are very similar.

Installing it on Linux was a breeze: sudo apt-get install quicksynergy

When I went to use OSX from a Linux Host, I found that Quick Synergy wasn’t able to configure the key mapping options that are needed to achieve keyboard zen. “No Problem”, I thought, “I’ll just edit the settings file that QuickSynergy Uses” – but it turns out that the changes are overwritten and my GUI dreams died with that issue.

I ended up finding the right key modifiers from Linux to OSX and reverted back to launching Synergy manually from the command line. synergys -f --config /path/to/synergy-config.conf

Its easy to setup aliases or other shortcuts but here is the magic sauce complete with key modifiers:

Have you ever been on a grails/groovy project that failed because of
- limitations/faults in the technology itself
- the grails/groovy learning curve

Here was my response:

Professionally, I have been a part of seven large Grails applications (+ more smaller endeavors). Of those seven, three of them were moving to grails from another framework, two due to failures with other teams/technologies and one was a Spring to Grails migration.

None of those seven failed because of Grails – in fact – the clients were very happy with the speed of development. Especially the rescue missions that had *very* tight timelines. At one startup we had a quite substantial Grails app that had a Flex front-end. When developing new stories, we found that the Flex front-end took ~80% of the time whereas the Grails components took 20% of our time and were very reasonable to test and maintain.

I’m not saying that we didn’t run into issues in learning curve or even issues/intricacies of Grails itself, but rather that we were able to make excellent progress and address those issues. The community and open source nature of both Groovy and Grails turned out to be extremely helpful when addressing technical challenges.

On one of the projects that had a very tight timeline most of the developers weren’t even Java developers, but had done some Django, PHP, RoR experience. The project was quite successful. We were able to accomplish more in less than a month then another firm was able to do in six months before declaring failure with another framework.

To add to my response, I would say that since testing is a priority and a first class citizen in Grails, it definitely added to the ongoing stability of the projects over time. The sections of code that were tested typically ended up being much more coherent and stable over the course of the projects life.

How would you respond?

java.net.BindException: Address already in use
at sun.nio.ch.Net.bind(Native Method)
at sun.nio.ch.ServerSocketChannelImpl.bind(ServerSocketChannelImpl.java:119)
at sun.nio.ch.ServerSocketAdaptor.bind(ServerSocketAdaptor.java:59)
...


It took me a second to figure out how locate the process, but it turned out to be a really easy task using the netstat command by executing ‘netstat -anp | grep 8080‘ (a = all interfaces, n = numeric, p = show pid)  This isn’t perfect since it will still show you anything with a pid of 8080 (the port number I was looking for), but its a great quick-and dirtry way to identify the process.  and once I’ve found it I can kill it with a ‘kill -9 <pid>‘

user@machine:~/$ netstat -anp | grep 8080
tcp6       0      0 :::8080                 :::*                    LISTEN      19884/java
user@machine:~/$ kill -9 19884

Which reminds me of this classic (please excuse the profanity):

ps:
Windows => netstat -ano
OSX => lsof -i -P

Playing with a multi-color LED

I found the Arduino IDE in package manager, and went to install it
sudo apt-get install arduino

Or Via Synaptic:

When I went to run it died with the following exception:

A little googling turned up the this bug. It basically describes that there is an issue with Ubuntu 10.10 + sun-java not including the jni directory in the classpath. There are several solutions, but basically this comment sums it up the best:

1) don’t use sun-java but default-jre instead. This bug is actually a known bug in sun-java. (bug #325506)
2) add export LD_LIBRARY_PATH=/usr/lib/jni to your .bashrc, the .deskop file (in /usr/share/desktop), or on the command line before starting arduino
3) add the new package (instructions are above).

I opted for #2 so far…  The idea flow for potential projects are unending.  I should try some CI build Lights/lamps next

I dual-boot OSX and Ubuntu, but primarily live, work and play in Ubuntu.  The only reasons that I boot up into OSX these days are to sync an iPhone, or more recently play some Steam Games (Go Steam for Releasing an OSX Client!  A Linux client shouldn’t be too far behind right?).  Windows is now relinquished to a VM for IE testing & other software that is stuck on that platform.

I’ve always been bothered by mice in OSX, they never felt good and didn’t handle as I had expected.  This is especially important when doing some intricate work with Photoshop, or in a computer game.  I don’t have much time to play games on the computer, but I’ve been able to get a few hours in recently and really noticed that the mouse acceleration was unusable/unreliable for fast-paced action…

A little googling turned up the “Mouse Acceleration Preference Pane” for Mac OSX by Christian Zuckschwerdt.

I downloaded version 1.1 and then had this preference pane is my System Preferences:

Which then lets me configure the acceleration for both the mouse and trackpad.

Even better, I could turn off acceleration all-together!

From Christian Zuckschwerdt’s site:

Extented mouse settings

“The Mouse Acceleration PrefPane is a GUI and startup item to Richard Bentley’s MouseFix.  You can set up extended speed parameters and acceleration curves for your mouse.”

Good work Christian!

I typically work running large web applications (Grails…) that typically use a lot of CPU power when running tests, launching the apps or refactorings within an IDE.  I like the ability to quickly adjust the CPU ‘govenor’ which governs how the CPU is utilized from a power/performance perspective (see CPU Frequency Scaling in Linux for more).

Ubuntu 9.10 (Karmic Koala) uses Gnome 2.28.1 which requires authorization to change the CPU Frequency (which makes sense).  Earlier versions did not require this authorization since it is new in Gnome 2.28.

With policykit-1 (also new in Ubuntu 9.10) you can grant yourself authorization for the cpufreq-applet based on a user or group by creating a policy file (at /var/lib/polkit-1/localauthority/50-local.d/org.gnome.cpufreqselector.pkla for example) that looks something like this:
[org.gnome.cpufreqselector]
Identity=unix-user:YourUser
Action=org.gnome.cpufreqselector
ResultAny=no
ResultInactive=no
ResultActive=yes

Thanks to this post/discussion for pointing this out   be sure you change YourUser to your username or group that you wish to have authorized.

I don’t know of a policy GUI that works with policykit-1 yet, so let me know if you find something.

I also replaced the icons in the /usr/share/pixmaps/cpufreq-applet/ with something I found on  gnome-look.org

I was hoping to spend some more time polishing this slide deck, but hopefully this presentation skeleton is helpful to the community as it is.  I am hoping to give this talk again at Gr8Conf in the USA this upcoming April and spend some more time creating more concrete and digestible examples.

I didn't have much time to figure it out but this is what I ended up finding out:

Grails 1.1.1 create-app generates a Config.groovy with this as the first few lines that would have told me what I need to know if I actually took the time to read it:
// locations to search for config files that get merged into the main config
// config files can either be Java properties files or ConfigSlurper scripts
//
// grails.config.locations = [ "classpath:${appName}-config.properties",
//                             "classpath:${appName}-config.groovy",
//                             "file:${userHome}/.grails/${appName}-config.properties",
//                             "file:${userHome}/.grails/${appName}-config.groovy"]
//
// if(System.properties["${appName}.config.location"]) {
//    grails.config.locations << "file:" + System.properties["${appName}.config.location"]
// }

So in my case all I had to do was put this in Config.groovy:
if (new File("${userHome}/.grails/${appName}-config.groovy").exists()){
    grails.config.locations = ["file:${userHome}/.grails/${appName}-config.groovy"]
}

which allowed me to override properties by doing something like this in my ~/.grails/${appName}-config.groovy :
username = "sa"
password = "sekret"
// environment specific settings
environments {
    development {
        dataSource {
            dbCreate = "create-drop" // use your imagination...
        }
    }
}

I also ran across a little gem that you can do the same type of config merging with your BuildConfig.groovy by implementing a ~/.grails/settings.groovy file. (yes its hardcoded rather than a config.locations property in BuildConfig.groovy – see BuildSettings.groovy for more)

Now I really wish there was a way to set defaults for things like server.port in the configs.

Thanks, Ted Naleid, Scott Vlaminck, and Robert Fischer for helping me find what I was looking for!