Thin Edge of the Wedge

Vodafone data charges

2008-04-15T23:40:00.004+01:00

I was out of the office so used my mobile as a modem.

Its a scandal! For 32mb I was charged €82 exclusive of VAT. This gives a grand total €99.25.

New residential broadband packages up to 15mbps from Smart Telecom

2008-03-11T21:56:00.003Z

Mobile data charges maybe excessive but I'm averaging down with my upgraded smart telecom package.

I now have 15mbps download, line rental, and voice package for €80 a month. All this with 1:1 contention, and no monthly download cap. Compare this to eircom's 12mbps business package: €169 per month exclusive of VAT (€205).

NB: Eircom's residential max is 4mbps at the moment.

Gulp - Data charges

2008-03-10T17:45:00.008Z

Update: I got my origanal sums wrong! I used 0.05 instead of 0.005 for calculating the per kilobyte charge. Its still expensive though. Thanks to Eddy for pointing it out. I have updated accordingly.

Sometimes I keep my 3 year old occupied in the car by letting him watch spiderman cartoons on Youtube via the Nokia N95.

If my calculations are correct and the prices listed of 2c per kb (VAT inc) up to 512 kb / 0.5c per kb over 512kb I'm in trouble.

That would mean I have to pay €12.80 for every 1mb downloaded!

kilobytes	price per kb	price
512	€0.02	€10.24
512	€0.005	€2.56
		€12.80

Ruby Ireland Slideshows

2008-02-16T13:04:00.002Z

I've added a sidebar to this blog which contains the Ruby Ireland presentations posted on slideshare. I'll try and keep adding the slideshows as they're given as long as the presenter is okay with it.

The latest one is SSH lightening presentation; which was given by David J Rice.

He also spoke about the start up he's involved with - Howcast. They launched last week and it looks fantastic.

XTech 2008 in Dublin

2008-01-27T22:50:00.000Z

There's an increasing amount of technology meetup and conferences happening in Dublin - Community and Commercial.

XTech 2008 is another in the coming months that looks interesting.

An Post Customer Services Helpline

2007-12-27T13:13:00.000Z

I just rang the number 1850 575859 to track a passport express application. Recorded message:

Welcome to An Post customer services. Our offices are now closed we will reopen on Tuesday the 10th of April.

Thats some christmas break :)

Block Facebook Beacon

2007-11-27T20:54:00.000Z

Here's how http://www.ideashower.com/blog/block-facebook-beacon/

Tesco UK offering PCs preloaded with Ubuntu

2007-10-26T09:14:00.000+01:00

Tesco UK is now offering PC's preloaded with Ubuntu

Intel Celeron PC for £139
AMD 3600 for £189

Active Merchant talk to Ruby Ireland

2007-10-08T20:42:00.000+01:00

I gave a talk at Ruby Ireland about Active Merchant, which is a ruby library for payment processing.

Slides

Here's the slides.

Unfortunately slideshare doesn't support animations so the callouts on the slides all show at once.

Sample Code

Click here for the sample application code

Ubuntu 7.10 - Gutsy Gibbon countdown

2007-10-05T02:02:00.000+01:00

Eircom Wireless Security Flaws

2007-10-01T13:45:00.001+01:00

I mentioned before about the lack of security in eircom's wifi solution. They use to have a single WEP key for all routers by default.

Now, the key is generated at installation but they've been really dumb in how they do it. James Galvin blogged about it and it was picked up by Damien. Paul Dolan put a page together to generate the key.

A new type of search

2007-09-21T09:59:00.000+01:00

When I'm stuck on the command line, trying to figure out regular expressions, or the correct syntax for a command I do the follow

Search Google
Select Page
Qualify Page is what I want and then bookmark using del.icio.us. The tags I use tend to be the same as search terms

What frequently happens, maybe days, maybe weeks later I'm doing the same thing. I repeat the process as its quicker than the 2 steps of seeing if I have something bookmarked already, and then searching.

The net effect is the value I obtain from del.icio.us is not as great as it could be. What I would like is a custom search indexes my del.icio.us links, then I could:

Search using my custom search
If any of my del.icio.us links match the terms in tags or content they appear first
Content from the web appears second

The benefit of this would be that pre qualified pages bookmarked previously would be a higher ranking. I suppose its a personal page ranking system.

Give me risk-free savings!!!

2007-09-17T09:25:00.000+01:00

Talk about ballsy, in yer face advertising. Thats exactly what Northern Rock had taking up the bottom third of page 5 of the Markets section in the Sunday Business Post.

On the same page as this article:
Sub-prime crisis rocks the banks

The Northern Rock collapse has focused attention on inter-bank loans and the vicious circle of credit that could send shockwaves through the mortgage market.

While on the previous page the following articles:
Sub-prime mortgages break Northern Rock

From ‘‘Northern Crock’’ to ‘‘Northern Rocky’’, it didn’t take long for the wags to add their bit to the meltdown of Britain’s fifth largest mortgage lender.

Credit crunch is set to hurt bank customers

Irish banks are poised to put the squeeze on customers to maintain their profits, writes David Clerkin.

Incredible!

Mission Impossible: Getting Legitimate email to Hotmail Part II

2007-09-14T09:12:00.000+01:00

Here's part I.

Microsoft have updated the process for submitting SPF/SenderID records. Although their guidelines suggest emailing senderid@microsoft.com if you do you will get the following response:

Thank you for contacting Sender ID support. We are in the process of providing a more robust support experience for the Sender ID Framework program. You can reach the support and submission form by submitting your request at our website: http://support.msn.com/default.aspx?productKey=senderid&mkt=en-us.

This is now an unmonitored alias - please use the above website to submit your request.

Thank you -
The Sender ID Framework Support Team

I stand by my original assertion that having to register an SenderID/SPF record is stupid.

Most spam filters check the SPF automatically by doing a DNS lookup on receipt of mail, but Microsoft insist that you register your domain with their SenderID program.

Obscure rails migration issue - value set to nil

2007-09-13T14:45:00.000+01:00

Migrations

Migrations in rails allow you to version your database. It works by checking the version of the database (using the schema_info table) and then running the appropriate ruby migrations in {RAILS_ROOT}/db/migrate.

As an example, if version number = 1 then

rake db:migrate

will execute all migrations named 002_{name}.rb and above in {RAILS_ROOT}/db/migrate.

You can also specify a version to migrate to using the VERSION parameter

rake db:migrate VERSION=X

This will execute all migrations named 002_{name}.rb and up to X_{name}.rb in {RAILS_ROOT}/db/migrate (assuming db version=1).

Obscure Issue

I have discovered a rather obscure issue in migrations, but only when multiple migrations are run at once.

Suppose our database is at VERSION=1, and we have the following 2 migrations.

oo2_add_column.rb

class AddColumn < ActiveRecord::Migration def self.up User.find(:all) do |u| #do stuff with users end add_column :users, :new_col, :string end def self.down end end

oo3_populate_column.rb

class PopulateColumn < ActiveRecord::Migration def self.up User.find(:all) do |u| u.new_col = "I'm new" u.save end end def self.down end end

The expected behaviour of running rake db:migrate when db is at VERSION=1 would be for users to have a new column and that column set to "I'm new" for every row.

However, that is not the case. The value of new_col will be nil.

How so? When ActiveRecord runs against a model it caches the column information so in migration 002 the User model column information is cached. Any subsequent saves in migration 003 will ignore the new column.

The obscurity of this is because that if each migration is run in isolation it works

rake db:migrate VERSION=2 rake db:migrate VERSION=3

This will result in correct behaviour as each migration is executed as a separate process and therefore the column cache is reloaded. This is typically how migrations are run in development.

Solution

The quick and dirty solution is to explicitly call User.reset_column_information after add_column. This resets all the cached information about columns, which will cause them to be reloaded on the next User model request.

Mashups using Google Mashup Editor and Yahoo pipes

2007-08-22T09:24:00.000+01:00

Introduction

Google have released a mashup editor, that makes it easy to create small functional web apps using their apis.

Yahoo have pipes which allows you to create, manipulate, and aggregate rss feeds.

The Mash Up

I want to create a mash up with the contents of the RTE news feed, hooked up with maps so the location of the news topic is pinpointed.

Use Yahoo Pipes

First I use yahoo pipes to suck in the RTE feed. Once complete I add a location extractor. This will scan the feed items and add a y:location element if it can successfully determine location from the content.

See the pipe output here.

Use Google Mashup Editor

I can then use google mashup editor (GME) to create a list of items from the Yahoo pipe.

The items when imported to GME get converted to atom format, and the y:location added by Yahoo pipes is added to the item's geo:lat and geo:long elements respectively.

Each item has an event handler when selected. When fired an item's geo location is parsed and used to pinpoint the location on the map.

You can see the end result here http://rtenewsgeocoded.googlemashups.com/ (currently if no location can be established the map point is random). If you select a news item the subsequent location will be pinpointed on the map.

Update: Use Yahoo Maps with pipes

Google maps with GME doesn't work well when only a few of the items are geocoded. It still plots them on the map at an arbitrary point.

I started looking at this from a number of angles, one being that the location is being determined from the feed content. When the feed contains only a title and one sentence summary, that makes it it increasingly difficult.

My first pass was to obtain the full content for the news article from another mashup tool: Dapper.

Once Dapper is setup to extra the full content, I need to loop through each item in the original feed and construct the URL to call the dapper service. The result contains the full news article content, which I use to replace the original item's description. This increases the complexity of the pipe.

You can see the updated pipe output as html here.

Once I have the detailed content it should mean a higher hit ratio for geo locations.

Pipe + Map output

Finally, rather than using Google Maps, the Yahoo map and pipes integration works better. See the final result here. The map tab shows these items that could be geocoded, while the list tab shows all items.

Embeddable Maps by Google

2007-08-21T15:53:00.001+01:00

If you use google maps you'll know they've always offered a link to this map option.

Now in a much anticipated move they have launched embeddable maps. It gives you the ability to embed maps in any web page with a snippet of HTML.

Here's an example:

View Larger Map

Excel Spreadsheet for getting geo locations

2007-08-16T10:37:00.000+01:00

Geocoding

Ireland was recently added to Google Maps Geocoder.

What I didn't know was that Yahoo already had geocoding for Ireland. In fact as far as simplistic interfaces Yahoo wins hands down. They have a simple rest interface for geocoding.

The parameters are listed on their site, but it means I can pass the following query and receive a response.

http://local.yahooapis.com/MapsService/V1/geocode?appid=YahooDemo&location=128+East+Wall+Road,Dublin+3,Ireland

Using in Excel

I started looking at using this from Excel and found that Jon Udell had already started something.

He used AutomateExcel.com and Juice Analytics as a starting point and then rolled his own.

I did similar using Jon's as the basis but reintroducing some aspects of the others. in particular removing the dependency on XML parsing.

The changes I have made are to include some extra columns. One, a very important one in my mind is a check link column. It autogenerates a link to Google Maps to the returned Geo code for manual confirmation. Also the precision, warning, and error columns are included to give back as much info as possible as the user.

I also elected to use the Yahoo parameter location which allows comma separated free text for the address, rather than the more specific params street,city,start,zip.

How it works

Simple really! Add one or more addresses from row 4 onwards, into columns 2-6 and hit Auto Geocode.

It only sends a request to Yahoo if Address 1 is filled in, and Auto Location is blank.

Before

After

Getting the spreadsheet

You can download it from here.

If decide to use it, please link back to this post.

Ireland added to Google Maps geocoder

2007-08-03T10:32:00.000+01:00

See the link here

http://googlemapsapi.blogspot.com/2007/08/looking-for-somewhere-in-india-hong.html

Enter an address without commas e.g. '128 East Wall Road Dublin 3 Ireland'

See an example

http://maps.google.com/maps?f=q&hl=en&geocode=&q=128+East+Wall+Road+Dublin+3+Ireland&sll=53.356421,-6.231713&sspn=0.016367,0.057335&ie=UTF8&z=15&iwloc=cent&om=1

Mongrel and Capistrano 2.0

2007-08-03T09:46:00.000+01:00

I use Capistrano, and Mongrel. Mongrel comes prepackaged with capistrano 1.4 recipes for stopping and starting, making it a cinch to roll a new version and stop and start your service.

Capistrano 2.0 has just been released, and mongrel hasn't yet been updated. The solution is to roll your own tasks for mongrel. You can do this by adding the following to your project's deploy.rb file.

Update: minor bug fixed in code. Thanks to funkdoobiest for pointing it out.

Update: mongrel_conf may need to be defined in your deploy.rb as meekish points out in the comments.

set :mongrel_conf, "#{deploy_to}/current/config/mongrel_cluster.yml"

namespace :deploy do namespace :mongrel do [ :stop, :start, :restart ].each do |t| desc "#{t.to_s.capitalize} the mongrel appserver" task t, :roles => :app do #invoke_command checks the use_sudo variable to determine how to run the mongrel_rails command invoke_command "mongrel_rails cluster::#{t.to_s} -C #{mongrel_conf}", :via => run_method end end end desc "Custom restart task for mongrel cluster" task :restart, :roles => :app, :except => { :no_release => true } do deploy.mongrel.restart end desc "Custom start task for mongrel cluster" task :start, :roles => :app do deploy.mongrel.start end desc "Custom stop task for mongrel cluster" task :stop, :roles => :app do deploy.mongrel.stop end end

Capistano already has tasks for deploy:restart, deploy:start, and deploy:stop. What the code above does is add the following tasks:

deploy:mongrel:start deploy:mongrel:restart deploy:mongrel:stop

It then overrides the default deploy:restart, deploy:start, and deploy:stop tasks to call the mongrel ones.

So now the following commands: stop, start and restart your mongrel servers respectively.

cap deploy:stop cap deploy:start cap deploy:restart

Testing RESTful route redirection in Ruby on Rails

2007-07-31T23:23:00.000+01:00

If you're doing functional testing with rails, you'll undoubtedly use follow_redirect - it allows you to follow the redirection for further tests.


  def test_something

    get :index



    assert_response :redirect



    follow_redirect



    assert_response :success

  end

However, the above code doesn't work when redirecting to a RESTful resource path. You'll get an error:

Exception: can't convert Symbol to String

This is because the URL helpers for restful resources return a string, where follow_redirect expects a hash.

The following code in your test_helper.rb file resolves this. It checks the type first to see if the normal follow_redirect should be called.

If not, it parses the path and calls the appropriate action thus following the redirection.


  alias_method_chain :follow_redirect, :restful_routes







  def follow_redirect_with_restful_routes

    #use the normal one unless its a string

    return follow_redirect_without_restful_routes unless @response.redirected_to.is_a?(String)    


    #okay we need to follow the redirect, but first parse the path

    url = URI.parse(@response.redirected_to)



    path = url.path



    extras = CGI.parse(url.query)



    #parse puts values into array so flatten

    extras.each do |key, value|

      extras[key] = value[0] if value.is_a?(Array) && value.length == 1

    end





    # Assume given controller

    request = ActionController::TestRequest.new({}, {}, nil)

    request.env["REQUEST_METHOD"] = "GET"

    equest.path = path



    redirected_controller = ActionController::Routing::Routes.recognize(request)



    if @controller.is_a?(redirected_controller)

      #then we can redirect, otherwise we can't'

      get request.path_parameters[:action], extras.symbolize_keys!

    else

       raise "Can't follow redirects outside of current controller (from #{@controller.controller_name} to #{redirected_controller})"

    end

  end

Dublin Codejam

2007-07-13T13:10:00.000+01:00

One of the exciting things happening with the Ruby Ireland group is Code Jam.

The principle was first proposed by Colm McHugh.

Oliver has setup a wiki page to discuss possible projects.

My ex-employer Oracle looking for rails developers

2007-07-09T17:56:00.000+01:00

If anymore evidence was needed that rails is going mainstream then here it is (via Ruby Inside).

Amazon EC2 - Paid AMIs

2007-07-06T09:12:00.000+01:00

Amazon Machine Instances (AMI) can be created with a specific software stack, registered and then run on EC2. Up until now they've relied on the community for public AMIs and I've used some of them. They have just announced paid AMIs where developers/companies can charge for prepackaged AMIs.

Free Ubuntu book online

2007-06-26T09:39:00.000+01:00

The Official Ubuntu book is available online for free.