theAdmin.org - Home

Open Source Bridge 2009

2009-06-25T04:02:00Z

I attended Open Source Bridge was last week, and have to say it was the best conference I've been to (I've only been to half a dozen). The variety of content kept the conference interesting and it was nice for me to finally meet others in the Portland tech community. I took a bunch of notes during the sessions and wanted to summarize them here.

Wednesday, June 17th

Welcome Session

The welcome session started with some statistics that I thought were interesting:

About 400 people attending
25% people are presenting
33% speakers are female

How to earn an open source living without taking investors or selling your soul

Brian Jamison from OpenSourcery talked about running a business that generates revenue from Open Source. He had a lot of good ideas, some of which I'm going to try out in Little Stream Software. My favorite idea was something we should all do more often:

"Say *No*, a lot

Specific customer

Sweat equity

Microsoft projects

Scope creep

Lowering price - you've lost if the customer asks"

Open Source tools for freelancers

Christie Koehler, a Portland freelance developer, went over several open source tools she used. She mentioned a small Python utility called timebook. It's a command line program that you can use to keep a running timer for tasks. I'm going to try it out for a few days to see if it can replace my paper timelog I keep.

How to build a successful open source software consulting company

Nate Aune from Jazkarta talked about starting an Open Source Consulting company. I already consider Little Stream Software an OS Consulting company, based on the work I do for Redmine but I did pick up few ideas on how to generate other streams of revenue (without selling my soul):

Custom development
Training
Support
Hosting

Information Security for the Open Source Business

Kevin Kenan gave a session on Information Security. Sadly, it was targeted at larger businesses than so many of the ideas weren't useful to me. A good idea I did get was that products should give their customers a guide showing how to configure the product securely along with the tools needed to test those security precautions.

Cluster Analysis - how to have fun in n dimensions

In Jesse Hallett's talk, showed how to do some Statistical analysis (Cluster analysis) to visually group data. It was a very interesting talk, though above my current math knowledge. He did have pretty animations though.

How to Work with the Government for fun and profit

I made a change at the last minute and decided to go to Deborah Bryant's session on how smaller businesses can win Government projects. I haven't had a lot of chances to work with larger organizations so I wanted to know how feasible it is to work with the local government.

Thursday, June 18th

Thursday Keynotes

The opening keynote for Thursday included a talk by the Portland Mayor, Sam Adams. He mentioned that Portland has a lot of funds for digital improvements but there isn't enough people trying to help. This might be a good opportunity for local tech businesses to pick up some work.

Ward Cunningham gave the next keynote about Innovations in Teamwork. One of his suggestion was that we try to work together more often, because that will lead to more insights and solutions than we have now. Based on that advise, I'm going to try to attend more local user groups.

Unit test your database

David Wheeler from PostgreSQL Experts Inc gave a talk about unit testing the PostgreSQL database. Even though Ruby on Rails doesn't use all of the features for each database, this talk had a few points that are great for writing unit tests against non-database code:

Test Driven Development is not for finding bugs, it for your own sanity and checking code consistently
If testing an interface is hard, then the function might need to be refactored. Hard to test means hard to use.
Tests are about as fast as your code but even then... they are much faster than debugging.

RubySpec: What does my Ruby do?

Brian Ford from Engine Yard gave a talk about RubySpec. I'm really excited about this project because he talked about some very easy ways to get involved and contribute. RubySpec is being used by many of the major Ruby implementations in order compare their implementation against the standard Ruby (MRI).

Configuration Management Panel

I went to this panel looking for a system that would help me automate server setup for myself and my customers. I was looking for something that was easy to use and that would make doing system upgrades simple. After hearing everyone speak, I'm going to take a look at puppet and automateit for my needs. The recording of this session just came online at blip.tv, I'd recommend watching it just to see the panel make offhand remarks at the other projects.

Bootstrapping Your Open Source Business

The GitHubbers never fail to entertain. Some notable quotes:

"Taking VC money is like marring a girl"
"I'm the CRO. Chief Revenue Officer"
"100,000 users on Twitter is the best integration test ever"

Git vs bzr smackdown

Selena Deckelmann and EmmaJane Hogbin paired on a session comparing git to bzr. bzr was my first exposure to a distributed version control system but git was the first one I really understood. From what Emma and the audience were saying, bzr is easier to learn and is good if you work with Windows. They also put up two sites to continue the battle: bzrvsgit.com gitvsbzr.com

Friday, June 19th

Friday was the unconference day. I didn't take many notes but I think Friday was still my favorite day just from the networking and discussions that came up. I'm going to have to attend more unconferences in Portland.

That's my write up of Open Source Bridge. I really enjoyed myself and would highly recommend attending next years conference. If anyone wants some more details about the notes I took, email me me and I'd be happy to send you a copy.

Eric.

On Safari - A web app to check SafariBooksOnline.com

2009-06-04T04:54:00Z

Earlier today I launched my very first Sinatra app, OnSafari. OnSafari is a tool to check if a book is available on Safari Books. I'm been a member of Safari Books for several years now and I'm constantly wondering if a book I'm looking at at a bookstore is on Safari or not. OnSafari will let me enter the book's ISBN number and will search Safari Books' website for the book.

Technology

My main goal with building OnSafari was to learn a few new technologies that I haven't had the opportunity to use yet.

Sinatra - Sinatra is a micro Ruby framework for building websites. It was very easy to pick up and learn how it works thanks to the documentation in the Sinatra Book. I'm definitely going to be using Sinatra again.
HAML - HAML is a templating language that uses white space to generate a HTML document. I didn't like the idea when I first heard about it but after using it for OnSafari, I'm starting to warm up to it. The templates are a lot easier to read but I did run into some unhelpful parsing errors. I think I'll give HAML another try on a future side project, but I'm not ready to recommend it for a customer project yet.
SASS - SASS is part of the HAML project and does basically the same thing but instead builds CSS files. I started to use SASS for the basic design. Once I integrated the final design though, I ended up dropping SASS in favor of vanilla CSS. Like HAML, I'd like try SASS out on a side project before I recommend it.
nokogiri - nokogiri is a HTML and XML parser. Since Safari Books doesn't have an API, I had to page scrape to "search" for books. I've already used Hpricot so I decided to give nokogiri a try. I had no problems with it, even when parsing Safari Books' JavaScript laden pages.
Heroku - Heroku is a hosting environment for Ruby Web application. I decided to try and host OnSafari on Heroku since they have been marketing themselves as the "Instant Ruby Platform". They really aren't joking, I was able to setup and deploy OnSafari in 15 minutes. That included making some code changes to accommodate their (lack of a) file system.

Summary

Overall, I'm very happy with the results of OnSafari. I learned a lot and was able to successfully build a site I've been wanting for awhile. I think my biggest surprise was how easy it is to host on Heroku. I'm now considering moving several other applications to their service, so I don't have to administer their servers anymore.

If you have a Safari Books account, I'd appreciate it if you would give OnSafari a try.

Eric

A Positive Change for the Rails Community - RailsBridge.org

2009-05-04T17:02:00Z

As any community grows, it will encounter challenges. Depending on how the members respond, the community will grow, shrink, or even be wiped out completely. Last week the Ruby on Rails community was presented with a challenge during a regional conference. Without getting into the details, the community mishandled the challenge and hurt many members of the community. Instead of complaining about the incident, I decided to team up with some other people in order to try and improve the community.

RailsBridge.org is born

After talking with Mike Gunderloy and a few other Rails community members, we started up RailsBridge. The website explains the purpose of the project perfectly:

The RailsBridge Mission: To create an inclusive and friendly Ruby on Rails community.

The RailsBridge Guidelines:

First, do no harm. Then, help where you can.

Bridge the gap from aspiring developer to contributing community member, through mentoring, teaching, and writing.

Reach out to individuals and groups who are underrepresented in the community.

Collaborate with other groups with similar goals.

The website has some more information about the project's goals, but my goal is to make the community more welcoming to new members. To do this, I'm going to be helping out with project Dana Jones is starting, Creating Courseware for Newcomers to Rails. If you would like to help make some positive change in the Ruby on Rails community, sign up for the group and let your voice be heard.

Eric

Redmineblog.com - The Official Redmine blog

2009-05-01T00:15:00Z

I'm happy to announce the launch of one of my Redmine side projects, the RedmineBlog.com. It will be the official blog for Redmine and will be another place the community can meet and topics discussed. From the About page:

We will be writing about various components of Redmine and it's community including:

We will try to make our content general enough to be used by anyone who currently runs Redmine or is evaluating Redmine, but we are targeting a few specific types of users to start:

Seasoned Redmine users who are looking for regular updates and upcoming news
New users who need detailed information on a specific topic
Ruby on Rails developers who are wanting to get into Redmine development

If you are interested in Redmine, I'd appreciate it if you would post any comments you have and subscribe to the RSS feed. Thanks.

Eric

Redmine Budget plugin v0.2.0 released (finally)

2009-04-27T23:07:00Z

I've just uploaded a new release of the Redmine Budget plugin. The Budget plugin is a plugin for Redmine to manage the set of deliverables for each project, automatically calculating key performance indicators. This is a long awaited release that will fix many of the outstanding bugs and incompatibilities that have cropped up.

Download

The plugin can be download from the Little Stream Software project or from GitHub.

Changes

You can see more details on the Changelog and Roadmap pages.

Help

If you need help, my Redmine bug tracker is open to the public and you are welcome to ask for help there.

If you are currently working on a Redmine plugin and need help or have an idea for a plugin you would like developed, please contact me. My company, Little Stream Software, specializes in the development of custom Redmine features and Redmine plugins.

Eric

Redmine Rate plugin released

2009-04-24T22:24:00Z

I've got a new Redmine plugin I'm Open Sourcing, the Redmine Rate plugin. This plugin will store billable rates for Users and will relate the amount of time worked to a specific amount. This was extracted out of the Budget plugin, once it was discovered that tracking historic rates was needed. Now if someone's billable rate changes, the plugin will adapt to that change without affecting past transactions.

Features

Track rates for a user based on
- Date Rate came into effect and
- the Project
Store historic rate amounts
Lock rates to preserve historic calculations
Rate.for API for other plugins
Integration with the Billing plugin
Integration with the Budget plugin

Getting the plugin

A copy of the plugin can be downloaded from Little Stream Software or from GitHub

Installation and Setup

This plugin will convert existing data in the Budget or Billing plugins. To prevent data loss, I recommend you read the README that is included with the release.

Usage

There are a couple of ways to use this plugin:

Enter new rate for a project

There are two ways to set rates for a Member of a Project.

Browse to the Project Settings page
Select the Members tab
Enter the rate for the Member and click the set Rate

Alternatively, Rates can be set in the User Administration panel

Browse to the Administration panel
Select Users
Select the specific user to add a rate for
Select the Membership tab and enter a rate for each project
Or, select the Rate History and enter a new rate in the form

Enter default rate for a user

A default rate is a user's Rate that doesn't correspond to a specific project. It can be set in the User Administration panel:

Browse to the Administration panel
Select Users
Select the specific user to add a rate for
Select the Rate History and enter a new rate in the form, keep the Project field set to Default Rate.

Lock a Rate

Currently this feature is only available through the Rate API. A Rate will become locked once a valid TimeEntry is assigned to the Rate. This is used to make sure historical amounts are not changed.

License

This plugin is licensed under the GNU GPL v2.

Project help

If you need help you can contact me or create an issue in the Bug Tracker.

Thanks

I would like to thank Shane and Peter, Inc for sponsoring this plugin. If you find it useful, send your appreciation his way.

Eric

How to modify core Redmine classes from a plugin

2009-04-13T18:15:00Z

I've been writing Redmine plugins since 2007 and one thing that stumped me was how to add new methods to Redmine's core classes and have them working in development. The standard Ruby on Rails way of including a module into the class works great except in development mode. Thanks to Thomas Löber, I found a way to overcome this error by making Ruby on Rails reload specific plugin classes.

Patching the core for fun and profit

I'm going to create a quick plugin that shows an example. Lets say you have a method called moo that you want to add to the Issue class. The standard Ruby on Rails plugin way would be to create a module and call Issue.send(:include).

# lib/my_moo_patch.rb
module MyMooPatch
  def self.included(base)
    base.send(:include, InstanceMethods)
  end

  module InstanceMethods
    def moo
      logger.info 'moo'
    end
  end    
end

# init.rb
require 'my_moo_patch'
Issue.send(:include, MyMooPatch)

This is a pretty standard case and would be used anytime your plugin needs to relate to the core classes. To illustrate this example and some errors that are caused, I added a call to @issue.moo on the Issue show page. Typically you would use the Redmine Hooks to have Redmine run your plugin's code.

Running the code

Running in the production environment, everything works fine:

Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 09:57:41) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}
moo
Rendering template within layouts/base
Rendering issues/show.rhtml
Completed in 1037ms (View: 873, DB: 31) | 200 OK [http://localhost/issues/1765]


Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 09:57:44) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}
moo
Rendering template within layouts/base
Rendering issues/show.rhtml
Completed in 257ms (View: 173, DB: 17) | 200 OK [http://localhost/issues/1765]

But in development, the first request will work but the second request will now throw an exception:

# First request
Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 09:59:32) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}
moo
Rendering template within layouts/base
Rendering issues/show.rhtml
Completed in 1285ms (View: 990, DB: 24) | 200 OK [http://localhost/issues/1765]

# Second request
Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 09:59:36) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}

NoMethodError (undefined method `moo' for #<Issue:0xb4ab3d94>):
    /vendor/rails/activerecord/lib/active_record/attribute_methods.rb:260:in `method_missing'
    /app/controllers/issues_controller.rb:98:in `show'
    <<snip>>

Rendering /home/edavis/dev/redmine/redmine-core/vendor/rails/actionpack/lib/action_controller/templates/rescues/layout.erb (internal_server_error)

Debugging the error

The NoMethodError is a standard error thrown by Ruby when a class doesn't have a method defined. But we defined Issue#moo in our plugin, so why is it not found? Lets take a look at what's happening in the Ruby debugger:

# First request
>> MyMooPatch.object_id
=> -621509448
>> @issue.class.object_id
=> -621657688
>> pp @issue.class.included_modules
[MyMooPatch::InstanceMethods,
 MyMooPatch,
 Redmine::Acts::ActivityProvider::InstanceMethods,
 Redmine::Acts::Event::InstanceMethods,
 Redmine::Acts::Searchable::InstanceMethods,
 Redmine::Acts::Watchable::InstanceMethods,
 Redmine::Acts::Customizable::InstanceMethods,
 Redmine::Acts::Attachable::InstanceMethods,
 Redmine::I18n,
 Kernel]
=> nil
>> @issue.class.included_modules.include?(MyMooPatch)
=> true

Nothing out of the ordinary here. Issue has MyMooPatch included and the request processes successfully.

# Second request
>> MyMooPatch.object_id
=> -621509448               # Same id as above
>> @issue.class.object_id
=> -630808728               # Different id from above.
>> pp @issue.class.included_modules
[Redmine::Acts::ActivityProvider::InstanceMethods,
 Redmine::Acts::Event::InstanceMethods,
 Redmine::Acts::Searchable::InstanceMethods,
 Redmine::Acts::Watchable::InstanceMethods,
 Redmine::Acts::Customizable::InstanceMethods,
 Redmine::Acts::Attachable::InstanceMethods,
 Redmine::I18n,
 Kernel]
=> nil
>> @issue.class.included_modules.include?(MyMooPatch)
=> false

But on the second request we notice something wrong. Issue no longer has MyMooPatch included and it also has a different class id from the last request. This means that something created a new Issue class object that is used instead of the original one. In development mode, Ruby on Rails reloads classes after each request so on the second request the Issue class was reloaded but our patch wasn't applied.

Wrapping our monkey-patch in a callback

Fortunately, Ruby on Rails provides a callback method we can use to add our module back, after a class is reloaded. It's called Dispatcher.to_prepare.

Add a preparation callback. Preparation callbacks are run before every request in development mode, and before the first request in production mode. (From: rails/actionpack/lib/action_controller/dispatcher.rb)

So we will need to change the way out module is included in Issue. Instead of just using the straight send, we need to wrap it in the Dispatcher.to_prepare method:

require 'dispatcher'
require 'my_moo_patch'

Dispatcher.to_prepare do
  Issue.send(:include, MyMooPatch)
end

Now if we restart the Ruby on Rails server in development mode, all the requests should be successful.

Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 10:42:58) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}
moo
Rendering template within layouts/base
Rendering issues/show.rhtml
Completed in 1382ms (View: 1056, DB: 26) | 200 OK [http://localhost/issues/1765]


Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 10:43:21) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}
moo
Rendering template within layouts/base
Rendering issues/show.rhtml
Completed in 1266ms (View: 951, DB: 26) | 200 OK [http://localhost/issues/1765]


Processing IssuesController#show (for 127.0.0.1 at 2009-04-13 10:43:25) [GET]
  Parameters: {"action"=>"show", "id"=>"1765", "controller"=>"issues"}
moo
Rendering template within layouts/base
Rendering issues/show.rhtml
Completed in 1210ms (View: 901, DB: 25) | 200 OK [http://localhost/issues/1765]

Success, our Issue is now mooing!

Summary

So if you write a Ruby on Rails or Redmine plugin that needs to patch the core, you should wrap your patches in Dispatcher.to_prepare so that they will work in development mode. I'll be converting all my plugins to use this pattern soon. If you would like to develop a patch for one yourself, fork one of my plugins on GitHub and sent me a pull request.

Eric

System Notification plugin v0.2.0 released

If I wanted to add a link example.com to the top menu how would I do that?

I've seen this asked several times and most responses are to modify Redmine's source. While this will work, it will make Redmine upgrades harder. The best solution is to create a small Redmine plugin and drop it into the vendor/plugins directory. Only one Ruby file is needed, plus any language translations.

# init.rb
Redmine::Plugin.register :redmine_static_link do
  name 'Redmine link'
  author 'Eric Davis of Little Stream Software'
  description 'Shows a link on the top menu'

  version '0.1.0'

  menu :top_menu, :example_link, "http://www.example.com"
end

You can see that all the init.rb does is to register a new plugin and add a :top_menu link to http://www.example.com. Once Redmine is reloaded, you should see the link on the top menu.

I've uploaded the full plugin, along with English language translations. Go ahead and download it and change the link url to what you need.

If you have any other Redmine or Redmine plugin questions that you would like me to write about, please ask them on my User Voice.

Eric

Use Hash#except to easily test Rails validations

2009-02-24T05:07:00Z

Ruby on Rails provides a large API for web development, including adding new methods to Ruby's base classes. One method I don't see used that often is Hash#except. From the Rails API documentation:

Return a hash that includes everything but the given keys. This is useful for limiting a set of parameters to everything but a few known toggles:
@person.update_attributes(params[:person].except(:admin))

A great use I found for it is to help test validation methods on Model objects. By defining a method that will return all the valid attributes on a model, you can easily exclude the ones you want to test with Hash#except.

I'm using RSpec in this example but the same idea works with Test::Unit or any other testing framework. I created a method valid_attributes that will return the attributes my Model needs to be valid. In this case my SystemNotification object needs a body, subject, and some users.

module SystemNotificationSpecHelper
  def valid_attributes
    user1 = mock_model(User, :mail => 'user1@example.com')
    user2 = mock_model(User, :mail => 'user2@example.com')
    return { 
      :body => 'a body',
      :subject => 'a subject line',
      :users => [user1, user2]
    }
  end
end

I used mocks for the users because I don't care about them, only that they are present. So to test if my object is valid, I can easily just pass in valid_attributes to my new method:

describe SystemNotification, "valid?" do
  include SystemNotificationSpecHelper

  it 'should be valid with the body, subject, and users' do
    system_notification = SystemNotification.new(valid_attributes)
    system_notification.valid?.should be_true
  end
end

Now I want to write some more specs to make sure if a SystemNotification is invalid if it's missing any of the required attribute. Since valid_attributes is a Hash, I can use Hash#except to remove each attribute individually and make sure the object is invalid for each case.

describe SystemNotification, "valid?" do
  include SystemNotificationSpecHelper

  it 'should be valid with the body, subject, and users' do
    system_notification = SystemNotification.new(valid_attributes)
    system_notification.valid?.should be_true
  end
  
  it 'should be invalid without a subject' do
    system_notification = SystemNotification.new(valid_attributes.except(:subject))
    system_notification.valid?.should be_false
  end
  
  it 'should be invalid without a body' do
    system_notification = SystemNotification.new(valid_attributes.except(:body))
    system_notification.valid?.should be_false
  end
  
  it 'should be invalid without any users' do
    system_notification = SystemNotification.new(valid_attributes.except(:users))
    system_notification.valid?.should be_false
  end
end

This is just a simple example but you can use Hash#except in many other places:

updating attributes like the Rail documentation sample
copying values from one object to another
extracting options from a argument

If you'd like some extra practice, refactor my specs to make sure SystemNotification#errors is populated based on what attribute failed. You can find the full code on GitHub, just send me a pull request when you're done.

Eric