Simulacre.org

Unblocking the Keyboard and Readline in EventMachine

2012-02-12T00:00:00Z

Reading from the keyboard without blocking under EventMachine is pretty simple. Just use the builtin EM.open_keyboard and give it a connection handler.

The Code Snippits Wiki for EventMachine gives an easy to follow example:

            
            require 'eventmachine'
            
            module MyKeyboardHandler
              def receive_data keystrokes
                puts "I received the following data from the keyboard: #{keystrokes}"
              end
            end
            
            EM.run {
               EM.open_keyboard(MyKeyboardHandler)
            }
            
            

If we want niceties like tab-completion, history, emacs edit mode, etc.,. you're out of luck. Normally you'd just use readline, but it will block. Maybe that's a good thing. Maybe you want the world to stop when asking the human questions. In some situations you don't want to stop the reactor loop though.

The best solution I've come up with is to replace $stdin with an EventMachine keyboard connection and use RbReadline.

# doesn't work with Stdlib Readline require 'rb-readline' require 'fiber' require 'eventmachine' module NbKeyboard def post_init @ostdin = $stdin $stdin = self @buffer = "" end def receive_data(d) @buffer << d @waiting && @waiting[:cnt] <= @buffer.length && @waiting[:fiber].resume end def read(cnt) if @buffer.length < cnt @waiting = {:cnt => cnt, :fiber => Fiber.current} Fiber.yield end data, @buffer = @buffer[0...cnt], @buffer[cnt..-1] data end def unbind $stdin = @ostdin end def method_missing(meth, *args, &blk) @ostdin.send(meth, *args, &blk) end end

RbReadline defaults to using $stdin for input, so when the NbKeyboard starts it makes $stdin point to itself. When RbReadline calls #read(1) on $stdin, NbKeyboard gives it back one character if it's available. Otherwise it pauses the current Fiber until the human types something in.

The Stlib Readline isn't supported because I haven't found a good way to get it to read from NbKeyboard rather than STDIN.

To verify that this works just run the demo:

#!/usr/bin/env ruby -wW1 require File.expandpath("../nb-keyboard", _FILE__) EM.run { spinner = EventMachine::PeriodicTimer.new(0.5) do $stderr.print "." end Fiber.new { EM.open_keyboard(NbKeyboard) do |kb| inp = Readline.readline("enter something ") $stderr.puts "you entered: #{inp.inspect}" EM::Timer.new(0.75) { EM.stop } end }.resume }

caleb@kiff % ruby demo.rb
            enter something ........ab.cd.ef.g.h.k.li..
            you entered: "abcdefghkli"
            ..
            caleb@kiff %

STDIN Blocking Note

Normally calling STDIN.read_nonblock(1) or STDIN.read(1) will not return until the user hits enter. The RbReadline and even standard Readline reset the terminal to not wait for the enter key to be pressed.

If you use EM.open_keyboard without RbReadline attached to it, your keyboard handler will not receive data until the user presses enter. You can get around that by using the stty program, or the rb-termios gem to reconfigure the terminal manually.

Pry Remote for EventMachine

2012-01-29T00:00:00Z

PryRemoteEm enables you to start instances of Pry in a running EventMachine program and connect to those Pry instances over a network or the Internet. Once connected you can interact with the internal state of the program.

It's based off of Mon-Ouie's pry-remote for DRb.

It adds user authentication and SSL support along with tab-completion and paging. It's compatble with MRI 1.9, or any other VM with support for Fibers and EventMachine.

It's up on Github, so fork away.

Installation

              
              gem install pry-remote-em

Usage

              
              require 'pry-remote-em/server'
              
              class Foo
                def initialize(x, y)
                  binding.remote_pry_em
                end
              end
              
              EM.run { Foo.new 10, 20 }

Running it will print out a message telling you Pry is waiting for a program to connect itself to it:

[pry-remote-em] listening for connections on pryem://localhost:6462/

You can then connect to the pry session using pry-remote-em:

)> stat Method Information: -- Name: initialize Owner: Foo Visibility: private Type: Bound Arity: 2 Method Signature: initialize(x, y) Source Location: (irb):5 [2] pry(#)> self => # [3] pry(#)> ls locals: _ _dir_ _ex_ _file_ _in_ _out_ _pry_ x y [4] pry(#)> x => 10 [5] pry(#)> x = 12 => 12 [6] pry(#)> exit [pry-remote-em] session terminated $ pry-remote-em [pry-remote-em] client connected to pryem://127.0.0.1:6462/ [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem [1] pry(#)> x => 12 [2] pry(#)> exit [pry-remote-em] session terminated 0x007fef1b22ac10>

              
              $ pry-remote-em
              [pry-remote-em] client connected to pryem://127.0.0.1:6462/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry(#)> stat 
              
              Method Information:
              --
              Name: initialize
              Owner: Foo
              Visibility: private
              Type: Bound
              Arity: 2
              Method Signature: initialize(x, y)
              Source Location: (irb):5
              
              [2] pry(#)> self
              => #
              
              [3] pry(#)> ls
              locals: _  _dir_  _ex_  _file_  _in_  _out_  _pry_  x  y
              [4] pry(#)> x
              => 10
              
              [5] pry(#)> x = 12
              => 12
              
              [6] pry(#)> exit
              [pry-remote-em] session terminated
              
              $ pry-remote-em
              [pry-remote-em] client connected to pryem://127.0.0.1:6462/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry(#)> x
              => 12
              
              [2] pry(#)> exit
              [pry-remote-em] session terminated
              
              

Features

Multiple Servers

It's easy to run more than one PryRemoteEm service on a single machine, or even in the same process. When you start the service via #remotepryem, just specify :auto as the port to use. The service will automatically take the next free port from 6462.

              
              require "pry-remote-em/server"
              
              os     = ObjectSpace.each_object
              expose = []
              while expose.length < 5
                o = os.next
                expose.push(o) unless o.frozen?
              end
              
              EM.run {
                expose.each {|o| o.remote_pry_em('localhost', :auto) }
              }
              
              

$ ruby test/auto-demo.rb
              [pry-remote-em] listening for connections on pryem://localhost:6462/
              [pry-remote-em] listening for connections on pryem://localhost:6463/
              [pry-remote-em] listening for connections on pryem://localhost:6464/
              [pry-remote-em] listening for connections on pryem://localhost:6465/
              [pry-remote-em] listening for connections on pryem://localhost:6466/

              
              $ pry-remote-em
              [pry-remote-em] client connected to pryem://127.0.0.1:6462/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry("pretty_print")>
              
              $ pry-remote-em  pryem://127.0.0.1:6463/
              [pry-remote-em] client connected to pryem://127.0.0.1:6463/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry("pack")>
              
              $ pry-remote-em  pryem://127.0.0.1:6464/
              [pry-remote-em] client connected to pryem://127.0.0.1:6464/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry("to_json")>
              
              $ pry-remote-em  pryem://127.0.0.1:6465/
              [pry-remote-em] client connected to pryem://127.0.0.1:6465/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry("to_json")>
              
              $ pry-remote-em  pryem://127.0.0.1:6466/
              [pry-remote-em] client connected to pryem://127.0.0.1:6466/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [1] pry(#)> 
              
              

TLS Encryption

When creating a server pass the :tls => true option to enable TLS.

              
              obj.remote_pry_em('localhost', :auto, :tls => true)

If you pass a Hash it will be used to configure the internal TLS handler.

              
              obj.remote_pry_em('localhost', :auto, :tls => {:private_key_file => '/tmp/server.key'})

See EventMachine::Connection#start_tls for the available options.

When the command line client connects to a TLS enabled server it will automatically use TLS mode even if the user didn't request it.

              
              $ pry-remote-em pryem://localhost:6462/
              [pry-remote-em] client connected to pryem://127.0.0.1:6462/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryems
              [pry-remote-em] negotiating TLS
              [pry-remote-em] TLS connection established
              [1] pry(#)> 
              
              

To always require a TLS connection give pry-remote-em a pryems URL. If the server doesn't support TLS the connection will be terminated.

              
              $ pry-remote-em pryems://localhost:6468/
              [pry-remote-em] client connected to pryem://127.0.0.1:6468/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryem
              [pry-remote-em] connection failed
              [pry-remote-em] server doesn't support required scheme "pryems"
              [pry-remote-em] session terminated
              
              

User Authentication

Server

If the service is started with the :auth option it will require all clients to authenticate on connect. The :auth option can be a Hash, proc or any object that responds to #call.

Auth with a Hash

              
              auth_hash = {'caleb' => 'crane', 'john' => 'lowski'}
              obj       = {:encoding => __ENCODING__, :weather => :cloudy}
              EM.run{
                obj.remote_pry_em('localhost', :auto, :tls => true, :auth => auth_hash)
              }
              
              

Auth with a lambda

              
              require 'net/ldap'
              ldap_anon = lambda do |user, pass|
                ldap = Net::LDAP.new :host => "10.0.0.1", :port => 389, :auth => {:method => :simple, :username => user, :password => pass}
                ldap.bind
              end
              obj       = {:encoding => __ENCODING__, :weather => :cloudy}
              EM.run{
                obj.remote_pry_em('localhost', :auto, :tls => true, :auth => ldap_anon)
              }
              
              

Auth with an object

              
              class Authenticator
                def initialize(db)
                  @db = db
                end
                def call(user, pass)
                  @db[user] && @db[user] == pass
                end
              end
              
              obj       = {:encoding => __ENCODING__, :weather => :cloudy}
              EM.run{
                obj.remote_pry_em('localhost', :auto, :tls => true, :auth => Authenticator.new(auth_hash))
              }
              
              

Client

The included command line client pry-remote-em can take a username and/or password as part of the url argument. If either a username or password is not supplied, but required, by the server, it will prompt for them.

              
              $ pry-remote-em pryems://localhost:6464/
              [pry-remote-em] client connected to pryem://127.0.0.1:6464/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryems
              [pry-remote-em] negotiating TLS
              [pry-remote-em] TLS connection established
              user: caleb
              caleb's password: *****
              [1] pry(#)> 
              
              

              
              $ pry-remote-em pryems://caleb@localhost:6464
              [pry-remote-em] client connected to pryem://127.0.0.1:6464/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryems
              [pry-remote-em] negotiating TLS
              [pry-remote-em] TLS connection established
              caleb's password: *****
              [1] pry(#)> 
              
              

              
              $ pry-remote-em pryems://caleb:crane@localhost:6464
              [pry-remote-em] client connected to pryem://127.0.0.1:6464/
              [pry-remote-em] remote is PryRemoteEm 0.4.0 pryems
              [pry-remote-em] negotiating TLS
              [pry-remote-em] TLS connection established
              [1] pry(#)> 
              
              

Tab Completion

Tab completion candidates will be retrieved from the server and presented on the client side.

)> key (^TAB ^TAB) key key? keys [1] pry(#)> keys => [:encoding]

              
              $ bin/pry-remote-em pryems:///
              [pry-remote-em] client connected to pryem://127.0.0.1:6462/
              [pry-remote-em] remote is PryRemoteEm 0.2.0 pryems
              [1] pry(#)> key (^TAB ^TAB)
              key   key?  keys
              [1] pry(#)> keys
              => [:encoding]
              
              

Paging

The standard Pry pager is supported through the included client.

)> ENV => {"COMMAND_MODE"=>"unix2003", "DISPLAY"=>"/tmp/launch-0EGhJW/org.x:0", "EDITOR"=>"mvim -f --nomru -c \"au VimLeave * !open -a Terminal\"", "GEM_HOME"=>"/Users/caleb/.rvm/gems/ruby-1.9.2-p290", "GEM_PATH"=> "/Users/caleb/.rvm/gems/ruby-1.9.2-p290:/Users/caleb/.rvm/gems/ruby-1.9.2-p290@global", "GREP_COLOR"=>"1;32", "GREP_OPTIONS"=>"--color=auto", "HOME"=>"/Users/caleb", "IRBRC"=>"/Users/caleb/.rvm/rubies/ruby-1.9.2-p290/.irbrc", "LC_CTYPE"=>"", "LOGNAME"=>"caleb", "LSCOLORS"=>"Gxfxcxdxbxegedabagacad", :

              
              [1] pry(#)> ENV
              => {"COMMAND_MODE"=>"unix2003",
               "DISPLAY"=>"/tmp/launch-0EGhJW/org.x:0",
               "EDITOR"=>"mvim -f --nomru -c \"au VimLeave * !open -a Terminal\"",
               "GEM_HOME"=>"/Users/caleb/.rvm/gems/ruby-1.9.2-p290",
               "GEM_PATH"=>
                "/Users/caleb/.rvm/gems/ruby-1.9.2-p290:/Users/caleb/.rvm/gems/ruby-1.9.2-p290@global",
               "GREP_COLOR"=>"1;32",
               "GREP_OPTIONS"=>"--color=auto",
               "HOME"=>"/Users/caleb",
               "IRBRC"=>"/Users/caleb/.rvm/rubies/ruby-1.9.2-p290/.irbrc",
               "LC_CTYPE"=>"",
               "LOGNAME"=>"caleb",
               "LSCOLORS"=>"Gxfxcxdxbxegedabagacad",
              :
              
              

Missing Features

AutoDiscovery/Broker ticket
HTTP Transport ticket
Shell Commands ticket
Vi mode editing - RbReadline doesn't support vi edit mode. I'm looking into contributing it. PryRemoteEm uses rb-readline because the STLIB version doesn't play nice with Fibers.
Ssh key based authentication

Issues

Please post any bug reports or feature requests on Github

Copyright

              Copyright (c) 2012 Caleb Crane
              
              Permission is hereby granted, free of charge, to any person obtaining
              a copy of this software and associated documentation files (the
              "Software"), to deal in the Software without restriction, including
              without limitation the rights to use, copy, modify, merge, publish,
              distribute, sublicense, and/or sell copies of the Software, and to
              permit persons to whom the Software is furnished to do so, subject to
              the following conditions:
              
              The above copyright notice and this permission notice shall be 
              included in all copies or substantial portions of the Software.
              
              THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
              EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
              MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
              NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
              LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
              OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
              WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Making GNU Screen Remember Splits

2012-01-11T00:00:00Z

GNU Screen is a great terminal multiplexer or simple window manager for CLI interfaces. Newer versions support spliting a single terminal window both horizontally and vertically. Each split window can run its own independent shell.

Screen also allows you to start programs detach from Screen, logout and later reattach to the same Screen after logging in again. Once reattached the programs will be safely where they were left.

Unfortunately whenever you detach from a Screen session, any split windows you setup in that session will be lost. The shells and programs running in the windows will be fine. You can get them back by recreating the split windows and connecting each one to a running shell.

That's a real pain though. Future versions of GNU Screen may remember the split window layouts after reattaching. Until then the only solution is to use a nested screen session.

Start one Screen session. Then immediately start a second session. Next reconfigure the escape sequence of the outer session to be Ctrl-o. Now use the standard escape sequence (Ctrl-a) to control the second - inner - session. When you're ready to detach use Ctrl-o d instead of Ctrl-a d. You can safely log out of the terminal now. When you want to get back just reattach to the first session.

Example

                
                # start one Screen session
                $ screen -S outer
                # start another Screen session in the first session
                $ screen -S inner
                # reconfigure the control key for the outer Screen
                C-a :escape ^Oo
                
                # start top in the inner Screen
                $ top
                # split the inner Screen and run vmstat in it
                C-a S
                C-a TAB
                C-a C
                $ vmstat 2
                
                # detach from the outer Screen
                C-o d
                
                # you can logout now
                
                # when ready, reattach to your Screen session
                $ screen -ls
                There are screens on:
                    15615.outer (Detached)
                    15692.inner (Attached)
                
                $ screen -r outer
                
                

Alternate ~/.screenrc

It's worth creating an alternate screenrc file with the escape keybinding for the outer screen session. Then whenever you start the outer session just point it at the alternate configuration file.

                
                $ cat ~/.screenrc.outer
                escape ^Oo
                $ screen -S outer -c ~/.screenrc.outer screen -S inner

References

Ssh Adapter for EventMachine

2012-01-04T00:00:00Z

Em-ssh is a net-ssh adapter for EventMachine. For the most part you can take any net-ssh code you have and run it in the EventMachine reactor.

Em-ssh is not associated with the Jamis Buck's net-ssh library. Please report any bugs with em-ssh to https://github.com/simulacre/em-ssh/issues

The code is on Github. The documentation is on rubydoc.

Installation

                  
                  gem install em-ssh

Synopsis

                  
                  EM.run do
                    EM::Ssh.start(host, user, :password => password) do |ssh|
                      # capture all stderr and stdout output from a remote process
                      ssh.exec!('uname -a').tap {|r| puts "\nuname: #{r}"}
                  
                      # capture only stdout matching a particular pattern
                      stdout = ""
                      ssh.exec!("ls -l /home") do |channel, stream, data|
                        stdout << data if stream == :stdout
                      end
                      puts "\n#{stdout}"
                  
                      # run multiple processes in parallel to completion
                      ssh.exec('ping -c 1 www.google.com')
                      ssh.exec('ping -c 1 www.yahoo.com')
                      ssh.exec('ping -c 1 www.rakuten.co.jp')
                  
                      #open a new channel and configure a minimal set of callbacks, then wait for the channel to finishes (closees).
                      channel = ssh.open_channel do |ch|
                        ch.exec "/usr/local/bin/ruby /path/to/file.rb" do |ch, success|
                          raise "could not execute command" unless success
                  
                          # "on_data" is called when the process writes something to stdout
                          ch.on_data do |c, data|
                            $stdout.print data
                          end
                  
                          # "on_extended_data" is called when the process writes something to stderr
                          ch.on_extended_data do |c, type, data|
                            $stderr.print data
                          end
                  
                          ch.on_close { puts "done!" }
                        end
                      end
                  
                      channel.wait
                  
                      ssh.close
                      EM.stop
                    end
                  end
                  
                  

See http://net-ssh.github.com/ssh/v2/api/index.html

Shell

Em-ssh provides an expect-like shell abstraction layer on top of net-ssh in EM::Ssh::Shell

Example

                  
                  require 'em-ssh/shell'
                  EM.run {
                    EM::Ssh::Shell.new(host, 'caleb', "") do |shell|
                      shell.should be_a(EventMachine::Ssh::Shell)
                      shell.wait_for(']$')
                      shell.send_and_wait('uname -a', ']$')
                      shell.wait_for(']$')
                      shell.send_and_wait('/sbin/ifconfig -a', ']$')
                      timer.cancel
                      EM.stop
                    end
                  }
                  
                  

Synchrony Example

                  
                  require 'em-ssh/shell'
                  EM.run {
                    Fiber.new {
                      shell = EM::Ssh::Shell.new(host, 'caleb', '')
                      shell.wait_for(']$')
                      shell.send_and_wait('sudo su -', 'password for caleb: ')
                      shell.send_and_wait('password', ']$')
                      output = shell.send_and_wait('/etc/init.d/openvpn restart', ']$')
                      # ...
                      shell.send_and_wait('exit', ']$')
                      shell.send_data('exit')
                    }.resume
                  }
                  
                  

Other Examples

See bin/em-ssh for an example of a basic replacement for system ssh.

See bin/em-ssh-shell for a more complex example usage of Shell.

Copyright

                  Copyright (c) 2011 Caleb Crane
                  
                  Permission is hereby granted, free of charge, to any person obtaining
                  a copy of this software and associated documentation files (the
                  "Software"), to deal in the Software without restriction, including
                  without limitation the rights to use, copy, modify, merge, publish,
                  distribute, sublicense, and/or sell copies of the Software, and to
                  permit persons to whom the Software is furnished to do so, subject to
                  the following conditions:
                  
                  The above copyright notice and this permission notice shall be
                  included in all copies or substantial portions of the Software.
                  
                  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
                  EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
                  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
                  NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
                  LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
                  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
                  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
                  
                  
                  Portions of this software are Copyright (c) 2008 Jamis Buck

Back in Time: Time Machine for Ubuntu Linux

2010-03-07T00:00:00Z

Back In Time is a great GUI that reproduces the most important OSX Time Machine features: simple graphical configuration and snapshot backups. The current version (0.9.26) is, unfortunately, broken if you want to back up an entire installation. There is a fix, but the developer has not released it yet. Applying the fix yourself is, fortunately, very easy.

Follow the directions below to install Back In Time for Ubuntu and backup your entire drive.

Install Back In Time

Add the following line to your apt sources file (/etc/apt/sources.list)

                    
                    deb http://le-web.org/repository stable main

Add the GPG key information:

                    
                    clc@lurr:~$ wget http://le-web.org/repository/le-web.key
                    clc@lurr:~$ sudo apt-key add le-web.key

Install the backintime application for Gnome

                    
                    clc@lurr:~$ sudo apt-get update
                    clc@lurr:~$ sudo apt-get install backintime-common backintime-gnome

Now the default installation of Back In Time should be installed. You can stop here, but if you want to back up your entire drive starting from the root ("/"), you’ll need to install a patch.

Download and install the patch

                    
                    clc@lurr:~$ wget http://launchpadlibrarian.net/35340108/snapshots.py.patch
                      --2010-03-07 12:28:39--  http://launchpadlibrarian.net/35340108/snapshots.py.patch
                      Resolving launchpadlibrarian.net... 91.189.89.228, 91.189.89.229
                      Connecting to launchpadlibrarian.net|91.189.89.228|:80... connected.
                      HTTP request sent, awaiting response... 200 OK
                      Length: 431 [text/plain]
                      Saving to: `snapshots.py.patch'
                    
                      100%[===================================================>] 431      --.-K/s   in 0s      
                    
                      2010-03-07 12:28:40 (13.9 MB/s) - `snapshots.py.patch' saved [431/431]
                    
                    clc@lurr:~$ sudo patch /usr/share/backintime/common/snapshots.py snapshots.py.patch
                      patching file /usr/share/backintime/common/snapshots.py
                    
                    

Easy as pie.

Configure Back In Time

Launch backintime-gnome (assuming your are not using KDE) and configure it.

                    
                    clc@lurr:~$ sudo backintime-gnome

Configure where you want the snapshots stored and how often.

Tell backintime which directories you want to save. I’m going for my entire drive starting from the root ("/").

Make sure to tell backintime not to backup special directories as well as the directory you use to store the snapshots. Mine’s under /mnt/backup, so excluding /mnt will get the job done.

Running as Root

If you want to backup your entire drive you’ll need backintime to run as root from root’s crontab. If you run backintime-gnome using sudo from the command line the config file for backintime won’t be saved in root’s .config directory though. When cron calls backintime it will look for the config and won’t find it, so you should create a link to the config file in root’s home directory.

                    
                    clc@lurr:~$ sudo ln -s ~/.config/backintime/config /home/root/.config/backintime/config

Debugging Back In Time

By default the output from Back In Time will be logged to /var/log/syslog. If your snapshots are not being saved as you expect check the syslog log.

                    
                    clc@lurr:~$ tail -f /var/log/syslog

I was stymied me for a bit when the cron daemon didn’t seem to be actually running backintime as it should. A quick tail of /var/log/auth.log pointed out that the root account’s password was expired.

                    
                    clc@lurr:~$ tail -f /var/log/auth.log

Optional: Configure Dedicated Logging

By default Back In Time will log its progress to /var/log/syslog. That of course means that every time it runs there will be changes to filesystem, therefore, every time it runs it will need to take a snapshot.

You can configure rsyslogd to push Back In Time’s logs to a dedicated file and then include that file in the exclude list.

Create and edit /etc/rsyslog.d/40-backintime.conf

                    
                    clc@lurr:~$ vim /etc/rsyslog.d/40-backintime.conf

Add the following lines

                    
                    if $programname == 'backintime' then /var/log/backintime.log
                    if $programname == 'backintime' then ~

Restart rsyslogd:

                    
                    clc@lurr:~$ sudo service rsyslog restart

Important Note

As of today, March 7th, 2010, the public release of Back In Time (v0.9.26) has not been updated to correctly handle taking a snapshot of an entire Ubuntu installation starting from the root directory ("/"). The instructions above are a quick guide for how to correct the issue. Check the Back In Time download page for the latest version of the application as the public release may have been updated.

Create a 2GB RAM Disk in OSX

2010-02-26T00:00:00Z

                      
                      diskutil erasevolume HFS+ "ramdisk" `hdiutil attach -nomount ram://4194304`

The only variable is the number of sectors. Calculate it by (size in MB) * 2048

Valentine Choco

2010-02-14T00:00:00Z

Yoshiko completely surprised me this Valentine’s Day with an insane Zoology Chocolate Chimpanzee and two cans of chocolate laced beer. Now I have to find a way to top her for White Day.

Figuring out how to eat him without melting chocolate all over my hands and face is only slightly harder than sitting through a Japanese driver’s license renewal class.

I guess I could cut him up with a knife, but that wouldn’t be nearly as much fun as eating him bit by bit.

Turn LI Elements into Links

2010-02-10T00:00:00Z

Horizontal menus of links look great and save space. Building them from UL elements is good practice and makes for easy CSS styline. The LI elements will often contain a link causing them to behave like buttons to trigger a page load.

Unfortunately, if the LI elements are setup with a lot of padding there will be a lot of dead space in the button that won't trigger the button's link.

You could make the entire button hot by wrapping the LI element in an anchor tag, but that wouldn't be valid: block elements should not be contained in inline elements.

Javascript & MooTools to the Rescue

Put the anchor inside the LI element and use javascript to assign an onClick event to the LI element.

                          
                          window.addEvent("domready", function(){
                            /** Make LIs into links if necessary. */
                            $$(".horiz-nav li").each(function(li){
                                var a = li.getFirst("a");
                                if(!a) return;
                                    li.addEvent("click", function(){
                                        window.location = a.get("href");
                                    })
                            });
                          });
                          
                          

Track Downloads and Exits Using Google Analytics and MooTools

2010-02-05T00:00:00Z

Last week I recorded 190+ medical engline terminology audio flashcards for Yoshiko and her classmates at Nihon University School of Medicine. They figured that it would be a lot easier to remember the words if they could hear a native speaker pronounce the words and phrases whenever they wanted. That's fine, but I wanted to know when they downloaded the MP3s.

By default Google Analytics won't automatically keep track of file downloads. It's pretty darn easy to do it manually though. Just add an onClick event to the anchor that calls:

                            
                            onClick="pageTracker._trackPageview('/download/mp3s.zip');"

That's pretty easy to do. I could even attach a similar event handler to all the links that lead off of my site, e.g., Twitter, in order to see how people left my site.

So it’s simple to add an event handler to the outbound links and downloadable files that I want to track, but it’s just such a pain to remember to add the tracking code every time I add a link. It also makes it a terrible chore if I ever decide to use Googles asynchronous tracking code. I’d have to go back and change syntax for every single link.

Well, 50 lines of code later I’ve got a MooTools module, ExitPoll, that tracks every outbound, or download link on a page. It even checks for the type of Analytics that the page is using: synchronous or asynchronous. That’ll make migrating to asynchronous tracking simple.

Anyway, you can download the latest version of ExitPoll from Github.

Just drop the module on your webserver, reference it, and then instantiate ExitPoll in your domready event handler.

window.addEvent("domready", function(){ new ExitPoll({ event : "click" }); });

                            
                            <script src="exitpoll.js" type="text/javascript"></script>
                            <script>
                            window.addEvent("domready", function(){
                               new ExitPoll({ event : "click" });
                            });
                            </script>
                            
                            

Take a look at the README for a full list of available options.

Setup a Content Delivery Network in 30 Minutes

2010-01-13T00:00:00Z

Using a Content Delivery Network will cut down on your hosting fees and improve user experience. Most CDNs are expensive. Building your own is horribly expensive and complicated. Google is cheap and easy, but in a good way.

Signup for Google App Engine
Create an app on GAE for your site.
Download the GAE SDK
Create a new application on your development machine
Create images, css, and js directories

Open up the app.yaml config file and create handles for the images, css, and js directories. The default_expiration tag is optional, but necessary if you want to use far-future expiration for long-term browser caching.

                              
                              application: simulacre-org
                              version: 1
                              runtime: python
                              api_version: 1
                              
                              default_expiration: "3650d 1h"
                              
                              handlers:
                              - url: /css
                                static_dir: css
                              
                              - url: /images
                                static_dir: images
                              
                              - url: /js
                                static_dir: js
                              
                              

Copy all your static content into the images, css and js directories
Deploy the app
Change all static content references on your site to point to http://[appname].appspot.com/

Optional, but a darn good idea – Setup a custom domain

Setup Google Apps for your domain if you haven't already
Login to your hosting environment and add a "static" subdomain CNAME that points to ghts.google.com., e.g., static.simualcre.org => ghs.google.com
Go back to the GAE Dashboard click on Application Settings then under Domain Setup click on Add Domain ...
Type in the name of your domain, click the add button and you'll be directed to the login for your Google Apps domain
Accept the Terms and Conditions
Under settings click Add new URL and enter the subdomain that you setup in stp 2, e.g., static
If you haven't already completed step 2, do it now then click I've completed these steps
Change all static content references on your site to point to your subdomain
You're done

Setting up the custom domain is a good idea because if you ever decide to stop using Google App Engine you only need to change the CNAME for the subdomain to point to another server that has the content.

Caveats & Tools

While Google App Engine does host your content on multiple servers it does not appear to serve content based on the geographical location of the requester.
Use Watchmouse to figure out if the GAE is serving content based on the requester's geographical location.
If you like Github give DryDrop a shot. It pulls your content from Github and pushes it to GAE

List of 500+ Look-Alike Kanji

2009-12-18T00:00:00Z

From the very start KanjiTastic has given you a quick way to retrieve kanji that look similar to the one that you are studying. All you have to do is click on the “Related Cards” link at the top of the card then select Similar Kanji.

That's all well and good, but not all kanji have a look-alike kanji connected with them yet. To make it easier to find look-alike kanji you now have access to our list of similar kanji.

We're up to 500+ plus kanji in the similar kanji list, so it's paginated and has a search box. You can enter either a page number or a kanji in the search box to quickly navigate to either a particular page, or a particular kanji.

Give it a try.

Twitter Cli Tools

2009-12-15T00:00:00Z

Just finished putting a little polish on a few Twitter command line tools that integrate nicely with GeekTool

twitter.php : Pull your friend_timeline from twitter. You can either setup your username and password directly in the script or pass them as command line options. Run ./twitter.php -h for a list of all the command line options.

twitterSearch.php : Search twitter for specific keyword(s). Try ./twitterSearch.php -s javascript,mootools for most recent posts that contain either the word javascript, or mootools.

twitterTimeline.php : Pull a specific user's timeline. I use it to monitor @ITJobsTweetUS without actually following them (they post so much that I loose track of my friend's posts).

Anyway, all the tools integrate well with GeekTool and support a few command line switches. Just run the tool with -h to get a rundown of the options. Only twitter.php requires you to provide a username and password. They all require that you have curl installed and in your path. If you're running OSX you're fine out-of-the-box. Most Linux distros will be fine too. Oh, you'll also need to have php in, or linked from /usr/bin/

https://github.com/simulacre/twitter-cli

Simulacre.org

Unblocking the Keyboard and Readline in EventMachine

STDIN Blocking Note

Pry Remote for EventMachine

Installation

Usage

Features

Multiple Servers

TLS Encryption

User Authentication

Server

Auth with a Hash

Auth with a lambda

Auth with an object

Client

Tab Completion

Paging

Missing Features

Issues

Copyright

Making GNU Screen Remember Splits

Example

Alternate ~/.screenrc

References

Ssh Adapter for EventMachine

Installation

Synopsis

Shell

Example

Synchrony Example

Other Examples

Copyright

Back in Time: Time Machine for Ubuntu Linux

Install Back In Time

Download and install the patch

Configure Back In Time

Running as Root

Debugging Back In Time

Optional: Configure Dedicated Logging

Important Note

Further Reading

Create a 2GB RAM Disk in OSX

Valentine Choco

Turn LI Elements into Links

Javascript & MooTools to the Rescue

Track Downloads and Exits Using Google Analytics and MooTools

Setup a Content Delivery Network in 30 Minutes

Optional, but a darn good idea – Setup a custom domain

Caveats & Tools

List of 500+ Look-Alike Kanji

Twitter Cli Tools