Alex Coles - Home

Crowd Control

2012-04-02T00:00:00-07:00

Crowd-funding meets open-source

Something makes me deeply uncomfortable about Yehuda Katz's (@wycats) recently announced crowd-funding effort for rails.app, and I have been trying to put my finger on it. To a certain extent, writing anything on the subject is a moot point. As of 2nd April, the target of $25,000 has been far exceeded. There is clearly interest (and money) in our industry for such a tool.

Nevertheless, I think Eric Hodel is conflating points in his latest blog post "On Community Funding of Open Source".

Over the last week, there have been those that have expressed their dissatisfaction with the way Yehuda lead refactoring of Rails 3. There were doubts about whether he was the right person to do the job. And there have, indeed, been tweets that border on the obnoxious and insulting. If we end up resorting to personal attacks, then Eric is right to call people out on that.

But, sceptic != hater. I am sceptical, and as such have found myself playing agent provocateur, retweeting various opposing views.

So why am I uneasy?

It boils down to the following three reasons:

1. No proof of concept. No sweat.

Yehuda is trying to appeal to a community of developers. We generally prefer to talk code, and not pour over Functional Requirement documents. A prototype, or proof of concept of some sort, would have been great.

This takes place in the context of a community that often espouses bootstrapping. The Travis CI developers sweated before asking for crowd-sourced funding. Prominent Rubyists like Amy Hoy champion bootstrapping whole-heartedly.

In the midst of this controversy, Jeremy McAnally (@jm) has already come up with railcar, a proof of concept that implements some parts of what Yehuda is proposing.

2. Oversight

When an open-source project comes about organically, through the sweat and initiative of its founder, there is usually no need for oversight. People will either use it, or not. A common adage of proponents of liberal licenses like the MIT (X11) is "I don't owe you fuck".

When a company sponsors the project, there is oversight in the sense that the company's profit imperative will usually keep the project somewhat on track.

But when a community is crowd-funding directly with a freelancer, how will this process work in practice? The scepticism does not arise from a mistrust of Yehuda. Rather, it arises from the concern about how individual developers can protect themselves against potential fall out when crowd-sourced projects go wrong.

rails.app will, inevitably, result in conflicting interests stemming from differing requirements. It could be anything from minutiae like the colour of the icon to the plugins bundled by default. For a foretaste of the problems that arise, you only have to look at the Rails and Bundler issue trackers. Their developers, on the other hand, can turn around and say "I don't owe you fuck". Yehuda won't be able to do this.

Whose interests will be prioritised? Which funder? The one who paid the most, or the one whose idea appears to make the most sense?

3. Money. Money. Money.

This is probably the least salient of the three main points, but I will mention it regardless. Here I could also easily be accused of jealously - perhaps my own fault for not monetising my own work as much as I should. But Yehuda is soliciting for work where he will be paid at a substantial rate – $275 per hour. In some parts of the world, this amount would fund an entire team.

… and a couple open questions

It is also important to consider what sort of precedent this will set for doing open-source. To what extent will this detract from funding projects that are already established? For instance, Rails Installer is planning a Mac version. If I were the maintainer of Ruby Installer, I imagine I would feel a bit resentful after ploughing hundreds of personal hours into a product (I am not the maintainer though – and knowing Luis Lavena (@luislavena), he is far more cool than me).

A final note

I cut my Ruby teeth on Merb, a spectacularly well-engineered Ruby MVC Framework. I have played with Janus and Thor. I use Bundler constantly. I am currently building products with Rails 3 and ember.js. I am a great fan of Yehuda's work, and enormously grateful for the contribution he has made to open-source. His work has enriched the Ruby community.

I also do not doubt the quality of his work. I would be happy to be half the programmer Yehuda is. For some, he may have made Rails 3 seem overly complex. For my needs at least, Yehuda made Rails 3 viable.

Do I doubt we need something like rails.app? At the beginning, I was not sure if there would be demand, but I will admit, my mind has changed after participating in the discussion on Twitter.

Yehuda deserves respect, and deserves to be paid a decent rate. I hope I have summed up some of the issues with the crowd-funding approach. If the community is taking on the role of investor, we must be more demanding. Yehuda is proposing a new model of funding open-source, and it is entirely right to criticise this process, in order to refine it.

As a footnote: this whole issue reminded me that the folks at Travis were due a little contribution from me. I was in the middle of an internal launch back when their crowd-funding started – read: living under a rock – and so it passed me by. I would encourage you to support an open-source project out there! Be it with cash donations or patches.

Bundler Pain Points

2012-03-20T00:00:00-07:00

In so many ways, Bundler is a godsend. Yet, like all gems (forgive the wordplay), it is not without a few rough edges.

While at the (absolutely fantastic) Wroclaw.rb last weekend, I finally took a moment to round up some of these issues into a lightning talk. Unfortunately, not having paced myself correctly I got cut off just before the final couple of slides. So I resolved to reformulate the content into a blog post. A further impetus was a recent gist by Sven Fuchs (der Travis CI- und Ruby I18n-Meister).

The aim of this article is to highlight a handful of — what I consider to be — the rough edges, and to proffer potential solutions/workarounds.

Perhaps surprisingly, one thing I am not going to talk about is the issue of performance (in other words, that infamous Fetching source index… wait). With the recent launch of Bundler 1.1 you should notice a big speedup. If you have not upgraded, please do. Bundler uses semantic versioning and this release should be fully API-compatible with Bundler 1.0.x.

Here is my run down of the top 3 Bundler pain points:

Lack of optional groups
Lack of overridable version dependencies
Not being able to declare a dependency more than once

1. Lack of optional groups

Use case

You are developing an application (as opposed to a library) and you want it to be usable and deployable 'out of the box.' You do not want your users to have to manually edit (or uncomment lines in) their Gemfile. You will find that this problem arises nearly always with database drivers.

Workaround

The current best workaround is something along these lines:

bundle install --without=postgres,sqlite3

In other words, rather than specifying groups to include, you have to provide a list of what you want to exclude.

gem 'pg',       groups: 'postgres'
gem 'sqlite3',  groups: 'sqlite3'
gem 'mysql',    groups: 'mysql'

A better fix

There is a proposal from @FooBarWidget and @mislav, which I find to be very pragmatic. In summary, this is what it would look like:

Installing optional groups,

bundle install --with=postgres

and in the Gemfile,

group :postgres, optional: true do
  gem 'pg'
end

I would encourage you to +1 the proposal!

2. Lack of overridable version dependencies

Use case

When a new version of Rails appears, it is frustrating to upgrade, only to discover that you cannot proceed until the plugins and libraries you depend upon are updated. In some cases, this is down to definite API incompatibilities, but in other cases, it might have been that the developer of the particular library was just too cautious in specifying the versions the library supports.

Workaround

The current, recommended workflow is to fork the particular library or plugin and specify it as a git dependency in your Gemfile.

A better fix

There is an issue open for this.

Thus, overriding dependencies when a particular gem gemspec is out of date:

gem 'sass-rails' do
  override 'railties', '4.0.0'
end

Alas, this is not something the Bundler team are currently considering adding in, and as such, there is no code. The Bundler team assert that the fork and git dependency workflow works well, but for my development team, it's turned into a maintenance nightmare.

My main argument is that a gemspec is not core application code, but instead metadata. As such, there should be mechanism for easily overriding this metadata – and that this mechanism should not be considered to be in the same light as monkey-patching.

I would also encourage you to comment on (or perhaps even +1) this issue!

3. Not being able to declare a dependency more than once

Use case

Bundler will not currently let you declare a dependency more than once. This means you cannot have different versions or sources for a dependency, even if you scope your dependencies using groups.

Not being able to declare a dependency more than once makes the following things painful:

testing against multiple versions of a resource (for example, in CI) (problem A)
development of componentised systems: (perhaps working against a git repository of a dependency, or an Engine in git) (problem B)

In fact, problem B provided my initial motivation for investigating the whole question. At Payango, we are currently building a new platform for prepaid VISA cards; our previous platform was built as a monolithic Rails application, but the new platform is built on the foundations of a Service-oriented Architecture. For each given product on our platform, we have multiple applications running. To share code, we are employing Rails engines and internal libraries extensively.

Of course, in a perfect world — and because we aim to keep concerns separate — our layers would be perfectly tested and respect their contracts. In this perfect world, we would never need to break down, debug or change views in our upstream library or Engine. Unfortunately — despite the best efforts and good design — we do sometimes need to work on a library and the host application simultaneously.

Workaround …

The concept is easy: the solution for both problems is to employ multiple Gemfiles.

… for problem A

Create more than one Gemfile, where you want to use different versions or a specify a tighter version than that defined in the .gemspec.

I have seen this solution used by Ripple, for example:

# Taken from:
# https://github.com/seancribbs/ripple/blob/master/Gemfile.rails31

gem 'activemodel', '~> 3.1.0'

instance_eval File.read(File.expand_path('../Gemfile', __FILE__))

where the .gemspec may contain,

gem.add_dependency "activemodel", [">= 3.0.0", "< 3.3.0"]

such an approach is useful for automated testing, particularly Continuous Integration. Travis CI allows you to specify multiple Gemfiles in your .travis.yml,

gemfile:
  - Gemfile.rails30
  - Gemfile.rails31
  - Gemfile.rails32

… for problem B

The solution lies in the environment variable BUNDLE_GEMFILE. First create a copy of your Gemfile as Gemfile.local and specify your path dependencies in your Gemfile.local, keeping your git or gem dependencies in your original Gemfile.

When it comes to working with the host application or library, just make sure BUNDLE_GEMFILE is set:

BUNDLE_GEMFILE=Gemfile.local bundle install
BUNDLE_GEMFILE=Gemfile.local rake spec
BUNDLE_GEMFILE=Gemfile.local rails server

of course you can wrap this into a shell function,

local_gemfile_on
local_gemfile_off

(See the gist for example functions you can add to your .bash_profile. You might also consider adding your current BUNDLE_GEMFILE to PS1)

Keeping more than one Gemfile is, of course, a big pain. So, use Rake to keep the copy in sync:

task :local_gemfile do
  gemfile = File.read('Gemfile')
  gemfile.gsub!(/git: ["']git@github\.com:payango\/([\w-]+)\.git["']/, 'path: \'../\1\'')
  File.open('Gemfile.local', 'w') { |file| file.write(gemfile) }
end

See the gist for the full tasks.

Two points to bear in mind when using this technique with Rails:

This will not handle code reloading: just because you do not have to bundle update each time you want to pull in the latest version of your library/Engine, it does not mean you will see all changes straight away. I have not yet worked out a solution for reloading Engines code on change.
If you are building on top of an older point release of Rails 3.x, you may need to update your config/boot.rb file. Because, older generated boot files overwrote the BUNDLE_GEMFILE environment variable, make sure your file looks like this.

A better fix

A better fix is on its way. Just as I was finishing this article, the following pull request from @josevalim was merged. I have yet to dig through the code, and it will take some time until the feature is released, but what is there already looks promising.

As a conclusion

Let us get this clear (and to end on a positive note):

The Ruby world before Bundler was a much more painful place. Manually installing Gems — with the ensuing Gem activation conflicts — was not fun. The Rails 2 way of dependency management, config.gem was buggy, at best.

Bundler has collectively spared us thousands of hours, and I hope to see it improved further.

What are your pain points?

Project Update: MarkdownJ

2009-11-25T00:00:00-08:00

The Web Dingus for MarkdownJ is back and deployed to Google App Engine. This is the first "application" I've deployed to GAE (I use inverted commas, as it is a pretty trivial application). The experience was relatively straightforward, although one issue I did encounter was that GAE's customised version of Jetty does not handle <jsp-file>-type servlets correctly (see issue 1365). A simple workaround was to rename the jsp file I wanted to map as a servlet to index.jsp.

My next step will be improving documentation and publishing the API docs somewhere. The MarkdownJ site is currently hosted on GitHub pages. I'm using Maven as build/project tool and have played a little with the possibility of publishing the generated docs to the gh-pages branch - it is possible (see Grancher for a Ruby tool that does this), but I don't find it particularly elegant.

Planned for early in the new year is a new MarkdownJ parser. Currently MarkdownJ relies on regular expressions, which doesn't provide for easy maintenance. Most importantly though, regular expressions don't really understand the Markdown code and don't provide for an intermediate representation of a Markdown document. An internal representation would make possible such things as multiple output formats (not just HTML, but also PDF and ODF) and pluggable converters to other markup languages. "Eclipse Mylyn WikiText", formerly known as TextileJ (eclipse.org must be hiring marketing managers from Redmond) provides this type of functionality for the Textile markup language.

Finally, since I'm mostly working in the Ruby language these days, I'd also like to see better support for JRuby. A thin wrapper Ruby-like API would be pretty easy to accomplish. It's particularly needed as the fast Markdown libraries for Ruby, BlueCloth and RDiscount, both rely on C native extensions, something JRuby does not support.

Getting Started with DataMapper on JRuby

2009-11-18T00:00:00-08:00

Underneath DataMapper (at least if you haven't bitten the NoSQL bullet just yet) lies DataObjects. DataObjects is an attempt to rewrite existing Ruby database drivers to conform to one, standard interface. If you're a recent arrival from the world of PHP or .Net, then the DO API will not look too disimilar to either PDO or ADO.net.

Late this spring, tucked away in the DataObjects' Release Notes for version 0.9.12, was mention of support for JRuby. Unfortunately the support that featured in 0.9.12 was incomplete at best. We needed a little longer to get things working well, but over the summer we made significant progress in supporting a large number of databases on the JRuby platform.

For a long while, I had a Gist up on Github illustrating the steps needed to get up and running with DataMapper on JRuby. I promised to follow up with a more detailed tutorial, but it's taken some time to get it published. In part, this has been down to the fact that going through the process of writing a tutorial has helped me uncover a plethora of bugs and inconsistencies, and I wanted to take the time to deal with them before hitting the metaphorical "publish" button. In any case, I hope this will be the first of many future Ruby postings.

For the purposes of this tutorial, I am using Merb as the MVC Framework, but you should also be able to use Rails, Camping, or Sinatra.

Many thanks to Dan Kubb (dkubb) for looking over this post and to George Adamson for actually having the patience to test out these steps (on Windows, of all platforms!)

Preflight

Install Java and the JDK

If you are running a recent version of Mac OS X, you should already have both the Java Runtime Environment and Java Development Kit (JDK). Apple supplies their own set of Java tools (although the Soylatte implementation is available if you want Java 6 on Tiger and Leopard 32-bit machines).

On Windows or a Linux distribution, you will likely already have the Java Runtime Environment, but unless you've been doing Java development, you may be without the JDK. The JDK is available from Sun's Java downloads page.

Install JRuby

We generally test against the latest stable version of JRuby. I've also tested against v1.2.0, but I'd recommend being on 1.3.1 or 1.4.0. Installing JRuby is as simple as downloading and decompressing the distributed .zip/.tar.gz file (or running the new .exe installer, if you're on Windows).

Verbose installation instructions for JRuby are now available on the Data Objects' wiki.

Configure your Path

Firstly, make sure JRuby is in your PATH. Typing jruby --version should give print out the version:

$ jruby --version
jruby 1.4.0RC2 (ruby 1.8.7 patchlevel 174) (2009-10-21 7e77f32) (Java HotSpot(TM) Client VM 1.6.0_15) [i386-java]

Because these instructions rely on installation from source, we'll also be compiling extensions that are written in Java. Hence make sure the JDK (and the Java compiler, javac) are in your PATH:

c:\Users\alexbcoles>javac -version
javac 1.6.0_16

On Windows for example, this would mean adding C:\Program Files\Java\jdk1.6.0_16\bin to PATH. Again, more verbose instructions can be found in the wiki.

Install DataMapper + Merb

A major requirement in developing DataObjects'/DataMapper's JDBC support was parity in the installation procedure. Installing on JRuby should be just as easy as for MRI / 1.9.

The good news is that if you want the latest published gems, you should be able to install them as follows:

jruby -S gem install do_sqlite3 dm-core dm-more merb

That's it! Unlike with ActiveRecord, you don't have to specify a separate JDBC variant.

However, for the purpose of these instructions and because our JRuby support is still developing at a rapid pace (for instance, our SQL Server support is only available in our Git repository), we'll cover installing the edge version of DataObjects / DataMapper.

Install Pre-Requisites

Note: Under Windows Vista/7, if you've used the JRuby installer and have your system gems directory in C:\Program Files\ then you may have to run cmd.exe with Administrator privileges (Start -> type cmd.exe, right click, Run as Administrator).

Note: On *NIX platforms, you may need to prefix commands with sudo, depending on where you installed jruby (unless, of course, you're using rvm).

Install Addressable and an implementation of JSON:

jruby -S gem install json_pure addressable

Install Jeweler (DataMapper and Merb projects are currently in the process of migrating their Rakefiles to use Jeweler):

jruby -S gem install jeweler

Install Extlib

Change to the directory where you keep your Code. (So ~/Dev, c:\code, c:\Documents and Settings\harry\Code, etc.)

We start by installing Extlib:

git clone git://github.com/datamapper/extlib.git
cd extlib
jruby -S rake install
cd ..

Note: If the rake install task fails under Windows, try again in two steps:

jruby -S rake package
jruby -S gem install pkg\extlib-0.9.14.gem

Extlib is a collection of extensions to core Ruby classes and is currently used by both the DataMapper and Merb projects. However, it'll go away in future versions as we start to take advantage of the new, modular ActiveSupport in Rails 3.0.

Install DataObjects

Then proceed to install DataObjects:

git clone git://github.com/datamapper/do.git
cd do

cd data_objects
jruby -S rake install
cd ..

Next, decide on the driver for the database vendor you wish to use. If you're installing from source, there will be three components to install:

do_jdbc: DataObjects' JDBC support library. This is a Java library, packaged as a Jar, and then wrapped up as a RubyGem.
the JDBC driver: the database vendor's driver using the JDBC API. Also a Java Jar, and in most cases, available wrapped up as RubyGems.
the actual DataObjects' driver: written in Java and relies on the above two libraries at runtime. Compiled against do_jdbc.

To install do_jdbc (the DataObjects' JDBC support library):

cd do_jdbc
jruby -S rake compile
jruby -S rake install
cd ..

To install the JDBC driver for the DataObjects driver, if you're using one of mysql, postgres, sqlite3 then we rely on ActiveRecord-JDBC to package the JDBC jar files as Gems for us. Installation should be a matter of:

jruby -S gem install jdbc-[DBNAME]
jdbc-derby
jdbc-h2
jdbc-hsqldb
jdbc-mysql
jdbc-postgres
jdbc-sqlite3

If you're on Oracle, then you need to place the Oracle JDBC driver ojdbc14.jar in your Java load path. Read Raimond's blog article for more instructions.

If you're on SQL Server, the ActiveRecord-JDBC has not yet packaged the jTDS JDBC driver as a RubyGem. However, we have this available in the DataObjects repository (you may need hoe first jruby -S gem install hoe):

cd jdbc_drivers/sqlserver
jruby -S rake install
cd ../..

Unless you've got your own project or have your mind set on a specific vendor's relational database, then follow this tutorial all the way through and install SQLite3 support (it means you won't have to do any configuration later!):

jruby -S gem install jdbc-sqlite3

Then you can proceed to install the DataObjects' driver:

cd do_[DBNAME]
jruby -S rake compile
jruby -S rake install
cd ../..

Install DataMapper

Installing DataMapper's core and more gems should then be very straight-forward:

git clone git://github.com/datamapper/dm-core.git
cd dm-core
jruby -S rake install
cd ..

git clone git://github.com/datamapper/dm-more.git
cd dm-more
jruby -S rake install
cd ..

Install Merb

The Merb maintainers just came out with a release (1.0.15) that should be compatible with DataMapper 0.10.x series. You should be able to install the gems as follows:

sudo jruby -S gem install merb-core merb-more

Create a Merb Project

Change to the project directory of your choice and use the merb-gen generator to create a "jump-start" Merb application. DataMapper is the default ORM for this generator type.

cd ~/Projects
jruby -S merb-gen app myapp
cd myapp

Before you run the application, make sure the dependency versions in config/dependencies.rb correspond with the "edge" versions you just installed.

Open config/dependencies.rb in your favourite text editor.

If the dependencies are out-of-date, then bump them appropriately:

merb_gems_version = "1.0.15"
dm_gems_version   = "~> 0.10"
do_gems_version   = "~> 0.10"

Note: The next major version of Merb, 1.1, will completely replace the dependency handling mechanism described here with Yehuda Katz's (wycats) Bundler. It's technically possible to use the Bundler with Merb 1.0.x series, but for concision, I won't elaborate on doing that here.

Run Merb

Before adding in any customisations of your own, check you're able to run the application.

Merb is built on Rack, and Rack needs a webserver. Webrick is a part of the Ruby standard library, so we could use that:

jruby -S merb -a webrick

(the -a flag specifies an adapter)

I encountered problems shutting down Webrick though, and had to kill it. Additionally, you'll probably want to use the server you'll end up deploying to, so let's try with Mongrel instead:

jruby -S gem install mongrel
jruby -S merb -V

(this time with -V to give us some verbose output)

I am personally a big fan of GlassFish for Java/JVM deployments. You can use something like Warbler to create a WAR file for deployment, but if you're coming from the Ruby world, the following is much simpler:

jruby -S gem install glassfish
glassfish

Add a Model

Here I start to get lazy. Bereft of ideas, and starting to think about needing to cook, I decided to create a simple Recipe tracker. While introducing some of the cooler (and defining) DataMapper features will have to wait until a future article, this simple child-parent model should be immediately familiar to anyone who's done basic web development.

Create a file ingredient.rb in app/models with the following contents:

class Ingredient
  include DataMapper::Resource

  property :id,           Serial
  property :name,         String, :length => 50, :nullable => false

  belongs_to :recipe, :nullable => true

end

and a file recipe.rb in app/models with the following contents:

class Recipe
  include DataMapper::Resource

  property :id,           Serial
  property :title,        String, :length => 100, :nullable => false
  property :instructions, Text
  property :cooking_mins, Integer, :min => 5, :max => 120

  has n, :ingredients

end

If you chose SQLite3 as the relational database you wanted to use earlier, then you don't need to change any database configuration. If you chose another database, then open config/database.yml and change appropriately. For PostgreSQL, or Microsoft SQL Server configurations might look like the following:

# PostgreSQL
development: &defaults
  adapter: postgres
  database: sample_development
  username: postgres
  password:
  host: localhost

# Microsoft SQL Server
development: &defaults
  adapter: sqlserver
  database: sample_development;instance=SQLEXPRESS;
  username: mswin
  password: doze
  host: sqlserver.localnet

Automigrate the database:

jruby -S rake db:automigrate

And run the application with one of the following commands:

jruby -S merb
glassfish

Add a UI

Here I'm sorry to say we're going to cheat (yet again!). We're not actually going to spend any time building a UI, but instead we'll rely on merb-admin, a project that automatically generates an admin front-end similar to Django, the Python web application framework.

jruby -S gem install builder
jruby -S gem install merb-admin -s http://gemcutter.org

In your app, add the following dependency to config/dependencies.rb:

dependency "merb-admin", "0.6.6"

Add the following route to config/router.rb (just after :merb_auth_slice_password):

slice(:merb_auth_slice_password, :name_prefix => nil, :path_prefix => "")
slice(:merb_admin, :path_prefix => "admin")

Then, run the following rake task:

jruby -S rake slices:merb-admin:install

Fire up your application, with one of the following commands:

jruby -S merb
glassfish

Visit http://localhost:4000/admin, or if you're running with glassfish, http://localhost:3000/admin.

You should have now have an application with a (very) simple interface for adding recipes and ingredients.

Uploaded with plasq's Skitch!

You can find the source code for this application on GitHub.

That's it for now! Please leave a comment and let me know what (if anything) was useful, so that I can focus my efforts for future articles.

UPDATE 1 Dec 2009: Grammar, spelling fixes. Added note on Jeweler.

The Hazards of Google Image Search

2008-08-14T00:00:00-07:00

http://news.bbc.co.uk/2/hi/uk_news/england/west_midlands/7560392.stm

Birmingham + skyline.

The Hazards of Online Translation

2008-08-05T00:00:00-07:00

http://adweek.blogs.com/adfreak/2008/07/then-well-grab.html

I have always been rather skeptical about online translation tools. This is a great illustration of what can go wrong for those who don't have any understanding at all of the target language.

A really simple way to integrate your Trac with GitHub

2008-05-15T00:00:00-07:00

Thanks to GitHub's API, there are definite possibilities for integrating Trac with GitHub. However, for right now the simplest way is to just provide a link. Thanks to customizable navigation introduced in Trac 0.11, this should now be really simple.

If you're using the web Administration screen, click through to Manage Plugins and from there, ensure trac.versioncontrol.web_ui.* modules are deselected, with the exception of the BrowserModule ( trac.versioncontrol.web_ui.browser). You can also enable/disable the appropriate plugins by editing your trac.ini file.

Then you'll need to remap the browser module link to GitHub. This time, this must be done by directly editing your trac.ini file. Open trac.ini in your favourite Editor and add the following section, if it does not already exist:

[mainnav]

and then the following line, changing the URL to match that of your GitHub project:

browser.href = http://github.com/myabc/github-trac/tree/master

And that should be it!

The best Hillary impression I have seen

2008-05-14T00:00:00-07:00

The best Hillary impression I have seen. Someone at Saturday Night Live needs to employ this woman.

Joining the bandwagon

2008-05-13T00:00:00-07:00

history | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head

140 git
103 ls
73 cd
55 mate
26 sake
26 mv
23 rake
8 autotest
5 gem
4 sudo

A prime explanation of subprime

2008-05-04T00:00:00-07:00

Update: Video no longer available.

State of Flux

2008-03-31T00:00:00-07:00

This site is in a state of flux while I work on various design improvements.

TheAtlantic.com | Andrew Sullivan

2008-02-18T00:00:00-08:00

http://andrewsullivan.theatlantic.com/

Good source of commentary, with snippets and links on the Presidential Primaries. An Obama supporter.

moche

2008-02-18T00:00:00-08:00

Yes, this site is rather ugly at the moment. That should change soon.

Keyword: Evil

2008-02-18T00:00:00-08:00

http://harpers.org/media/slideshow/annot/2008-03/index.html

There is an great feature in this month's Harper's Magazine on Google's Data Centre plans for a small community on the banks of the Columbia River in Oregon. I think its really easy to see technology as something invisible: after all, thats why Google searches, Gmail and YouTube are so popular.

But this technology is not really invisible. It needs power. The brain power is more apparent: Google is actively expanding and a huge number of "star engineers" have joined its ranks. The less apparent, oft-neglected power is electricity.

While it may be out-of-sight (and off-site) for most of us, its dangerous for us to let this important fact slip out-of-mind. The immense power that Google affords us relies on huge amounts of processing power, and in turn, huge amounts of electricity. The article is quite right to point out that Google has now become a heavy industry of the 21st century.

While I don't know whether its more efficient, in terms of power consumption, for individuals, small business and communities to do things themselves (such as running email servers, etc.), or for everything to be consolidated into Google's huge farms of servers, I do know that it is important not to not consider the efficiency and environmental record of this new heavy industry.

Just words?

2008-02-18T00:00:00-08:00

Colours

2008-02-18T00:00:00-08:00

This blog is currently undergoing transformation. Change colours.

Changes behind the scenes

2008-02-18T00:00:00-08:00

I do realise I spend more time blogging about blogging, than actually blogging.

I spent a couple hours moving away from WordPress, which was the blogging software this site has used since its inception a couple years ago. I moved to Chyrp, a new lightweight blogging engine written in PHP. Admittedly, some of this is the thrill of the new. I do like to be current. And some of this was procrastination from other tasks at hand.

I had become frustrated with WordPress. That's not to say its not very powerful, nor that I wouldn't still recommend it to clients in certain instances. But for me, it doesn't work. Sitting down and using the supplied administrative application as a writing tool has never worked for me. It doesn't "mesh" with my brain (for want of a better word). As a solution, I've tried various external clients (ecto, Flock), but they have never really worked for me either.

Much the same as that bête noire of writers out there, Microsoft Word. If there could be a prescription for drawing a blank, de-voiding the mind of all ideas and creativity, then opening a Microsoft Word window would have to be it.

Writing tools are important. I am not a "writer" by any stretch of the imagination, but I do want to get back into the action of writing (see also this post)-- perhaps in pursuit of some generalised catharsis.

For me, the fewer distractions the better. The next best thing to a piece of paper (or even better, a Clairefontaine notebook), is a simple text editor (such as TextMate using with a fixed-width typeface.

Don't get me wrong. I don't necessarily long the days of typewritten documents, or everything in Courrier New. But the WYSIWYG (what-you-see-is-what-you-get) of Word, other word-processors, and even that pseudo-wordprocessor toolbar that you get in GMail these days, does distract.

I loved Desktop Publishing. I love using Illustrator. I am fascinated by good design. However, I would never begin to start composing anything of any considerable length directly into those programs. In all but a few cases (perhaps composing visual poetry, which alas, I have never tried), it would be an immense distraction. My mind starts to think in terms of bold, italic, size and kerning. My mind loses its focus on the words themselves.

So back to blogging. So far (admittedly its only been a handful of posts), Chyrp seems to work better for me. Chyrp follows the "tumblelog" way of doing things, which for the uninitiated (as I was, about 3 hours ago), is a short form style of blogging, with plenty of links, video and stuff like that. Great for the easily distracted, like me. Although, with any luck, there will be some longer prose here too.

I set it up to work with Markdown, a formatting syntax that is pretty easy and isn't verbose like adding HTML itself. If a particular word really warrants being emboldened or italicised then I can do so. If I really want.

The slogan of WordPress is "code is poetry". I am not sure what genre of poetry they are writing, but to me, reading there lines of code is like reading Dadist poetry. There are a mixture of coding styles (object-oriented, procedural) and lots of hacks to get account for the fact that PHP, the language its written in, can be rough around the edges (I'd be happy to justify that, if any one really cares). On the outside, it looks elegant, but navigating its innards is difficult. There are a tonne of plugins and themes available for download, but many of those I've tried, I have found to be poorly designed. In some sense, it suffers from its own success, in that popularity has brought a tonne of developers who want to use the software. But there is no quality control, and as a writing tool, it hasn't worked out.

Some of my problems were also design related. Chyrp should be easier for me to customise, so look out for some design changes over the next few days.

Update: Chyrp has, by no means, been a panacea. In fact, I've taken several months just to fix what needs fixing. A new version is out, and I hope that some of the problems I was experiencing have finally been rectified.

Americans Away from Home

2008-02-18T00:00:00-08:00

http://www.americans-away-from-home.com/

A website written by an American expat following the course of the U.S. Presidential Primaries.

Battle of the Bands

2008-02-15T00:00:00-08:00

Do we have a winner?

Life on Mars

2008-02-14T00:00:00-08:00

I am using <a href="http://www.red-sweater.com/marsedit/">MarsEdit</a>. Perhaps this will make it easier for me to publish when I feel that sudden writing impulse.

Bienvenue et Welcome

2008-02-14T00:00:00-08:00

Welcome to my new Portfolio site - mon book en ligne. I have been meaning to get this site up for quite some time, but I resolved to get something up, so here it is. I'll be working on it bit-by-bit, but there have been a few false starts.

An original version of this message appeared on a WordPress-powered blog a few weeks ago, but I have since moved over the the Apache Roller weblog software. I'll explain why I did sometime soon.

This is not the extent of this site. There is an awesome Ruby-on-Rails-powered application that I'll be launching soon to show off the work I've done over the last few years (at least that which isn't under NDA), but for now please enjoy this blog. I'll use the blog space to detail some of the projects I'm working on (including some open-source projects), as well as to share various bits of information I've learned (or gleaned) that have helped me to get up and running developing websites and applications.

Of course, I could have mixed this all in with my ~~personal blog~~ (update: content now merged back into alexbcoles.com). Perhaps I'm being a little self-indulgent by having two blogs (neither of which are updated quite as often as they should be), but I believe the audiences are different. The innards of PHP, Java or Ruby configuration don't (and shouldn't) interest everyone.

Here it is. Thanks for reading.

Interlude

2008-02-03T00:00:00-08:00

New Year, New Start and Finish

2008-01-25T00:00:00-08:00

I have a constant stream of ideas, but I do have to admit to finding it hard to finish things. My lack of content on this blog is rather indicative of some of the difficulties, not with picking up the (metaphorical) pen, but with laying it to rest, adding a full stop, and closing the paragraph.

While I've liked the idea of having a blog, its not something I have really put to use. A blog, it seems, is something is pretty much expected for those making their living in the technology industry. In my case though I am told that this blog is bordering on 'pitiful' for its lack of content (the words, nonetheless, of one of my closest friends). Instead it has become more of a 'metablog' with me blogging on blogging, and offering my excuses for not writing more often.

Up until a few weeks ago my WordPress 'Draft' folder was littered with the remnants of half-finished writings that had been created over the last 2 years or so, but never finished. In a fit of desire for some closure, I went through and deleted the entirety of these half-baked works... (Parenthetically, somehow my observations on the student politics, dirty student politics, and minor scandal, no longer seem relevant to me.)

I am still suffering from the same condition that I first encountered in my early teens. I start the first paragraph with such zeal and optimism. Ideas pouring over. Then my ambitions get the better of me, and the work becomes unsustainable. I remember my creative writing in secondary school would show this. Beginning. Middle. No End. I was once so flippant as to add 'to be continued' on an end of term essay, which managed to earn me not only a letter mark far down the Greek alphabet, but an equally flippant response from my teacher. Well, this year I must be better.

Update 2

2007-10-29T00:00:00-07:00

I am in Paris, France. More soon.

Additional Blog

2007-10-29T00:00:00-07:00

I have set up a separate site and blog for work-related interests. Find it at ~~alexcolesportfolio.com/blog~~ (update: content now merged back into alexbcoles.com).

Update

2007-03-10T00:00:00-08:00

Once in a very occasional while a visitor stumbles across this excuse-for-a-blog. And when they do, they feel inclined to chastise me for lack of content. So, I thought I'd let any occassional visitors know what I am up to.

After completing university last year, I came over to the US for the summer. After completing most of a couple freelance jobs, I stopped in New York. I hadn't planned to stay, but somehow I ended up doing so... after several weeks of trying half-heartedly to look for work, then a couple weeks of serious "I'm broke I better get a job" looking, three jobs came along at once. I took a job with a technology startup company, which I thought would be the most interesting - and most challenging.

I've now been in New York coming up to six months, and in my current job for 3½ months. There are plenty of things I miss about being in London, but its great to have the challenge of being in a new place.

Blog back up and running...

2007-01-18T00:00:00-08:00

Somehow I managed to delete the blog, and its postings. But it is up again, once more. I lost my last posting, but it wasn't anything prolific (unlike most of what I have to say ;) )

There is stuff to come...

2006-03-27T00:00:00-08:00

This site is being worked on!

My project over the holidays is to get my Portfolio online. I looked around on the web and there isn't any software out there to handle creating a portfolio, so I'm working on creating something from scratch.

Writing

2006-03-13T00:00:00-08:00

I received a friendly reminder to update my blog the other day.

I still haven't got the hang of it. It's partly because I don't really think I lead an eventful enough life to talk about it in great detail, and it's partly that I am not sure a blog deserves a great deal of my time. I could (and may still) start commenting on issues of "current importance", but that would be very time consuming.

I am in the midst of writing essays. I have 8 or 9 essays to write in the next 6 weeks, and then that will be it. Unless I prevaricate and do a Master's, never again will I have to write an essay.

But I digress.. I have the same problem with writing essays as I do with writing blogs. The first few postings (which were really for test purposes) were an exception. I seek perfection. I don't want to output more dross, prosaic banalities. If I'm going to have a blog I feel I should write something gripping.

In the window in which I am typing there are five draft articles listed. They're all unfinished, because I don't yet think they're suitable for general consumption.

On my desk there are three draft essays. They're also also unfinished, because I don't yet think they offer any originality of thought.

I would say it was Writer's Block, but then my levels of pretension would soar. I am, most likely, not a Writer.

Quick test message

2006-01-21T00:00:00-08:00

Quick test message to test ecto, the new blog editing software I am trying out.

Invites for Windows Live Messenger

2006-01-21T00:00:00-08:00

Microsoft have rebranded the next version of MSN Messenger as Windows Live Messenger. Information is at Windows Live Ideas.

Much like with the birth of Google's GMail, download is currently by invitation only. I have a few invitations to spare, so if you're interested, leave a message on this blog.

Using Flock

2005-12-29T00:00:00-08:00

For the first two posts on this blog I used the WordPress editor through my browser. Despite following certain Internet trends closely, I am not at all au fait with blogging. I did remember though that there were some external editors available. Trying them out might encourage me to blog more often.

A couple months ago I downloaded Flock. I had read a couple articles about it, and I wanted to see what all the hype was about. My first impression was that I liked the look and feel a lot, but because I didn't feel any need for the 'value added' features (which is the raison d'être of this browser) I haven't opened it since my first install.

After doing a search to look for external blog editors, I recalled that Flock has an editing feature. Its this editing feature that I am using at the moment to write this blog entry. I don't really know what a blog editor should look like though, so I'd welcome some recommendations from the four or five of you that currently read my blog! It doesn't have a spell checker (although that doesn't necessarily bother me that much). Forgive the 'woolly' terms, but it does feel 'snappy' and 'comfortable' as a writing tool. That's more than I can say about the second most used application on my Mac desktop (Word)!

Where in the World?

2005-12-08T00:00:00-08:00

I'm always on the look out for ways to waste time online. KLM make up for their mediocre in-flight service by providing a wonderful public service that plots where you've been in the world. Use to boast to your friends how well travelled you are!

I suppose this also helps me prioritise where I want to go next. For various reasons I spend most of my time between London and the States, but I really want to spend more time in S. America. Brazil and Argentina have to be top of my list. I wish I had the money, but even more, I wish I had the time!

Update: The map link is broken, and thus, removed.

First Posting

2005-12-08T00:00:00-08:00

After various failed attempts at blogging (including two when I was a Sabbatical Officer at UCL Union), I'll now try again. I am not sure I'll say too much about myself, as I don't really think I'm that interesting! But I'll link to things that are interesting when I find them. Blogging is another great distraction in life... something that'll lead me down the road to a third-class degree.