Mike Bowler

I've been using Eclipse Ganymede for a couple of days now and here are a couple of first impressions.

• Overall, everything is faster. There's a snappiness to the UI compared to the last release.
• There's a subversion client (subversive) bundled with the package which would be good news except that I couldn't get it to work. Apparently, it's deliberately broken since they're not allowed to ship some needed component so you need to download some extra piece. I tried for a bit and while I think I downloaded the right part, I never did get it working. Finally, I removed subversive and went back to subclipse and it's working fine for me.
• The dynamic language support is now built-in although ruby support is still an extra step in the installation.
• Syntax highlighting in ruby has taken a step backwards - it gets really confused by quotes in places other than strings (in regular expressions for example) and then makes a mess of the syntax highlighting after that. I've been able to work around this by putting comments with end quotes in them but this is clearly not ideal. I hope they get this one fixed soon.

The slides for the talk I did last night at the Rails Project Night are on SlideShare at http://www.slideshare.net/mbowler/inside-enumerable

One of the big milestones in a ruby implementation is when it's able to first run a simple rails application. Until a couple of weeks ago, there were only two ruby implementations that were able to do this - the C reference implementation (Matz's ruby) and JRuby.

Today, there are four implementations that have hit this milestone. Rubinius and Iron Ruby have both just announced that they're able to run a simple rails app. This is an impressive milestone for both projects - congratulations to everyone working on those.

I've put the slides from last nights talk up on SlideShare. This is the first time I've used SlideShare so if you have any trouble accessing the slides, let me know.

The slides for last night's talk at TSOT are here. The talk was a walkthrough of the code in Date.once()

Note that I've stripped out the background in order to make this a much smaller download. It doesn't look as pretty but it's a fraction of the original size.

HtmlUnit 2.0 has been released. Although I started this project back in 2002, I haven't been actively involved in a long time. I'm really pleased to see how far it's progressed since I let go.

Marc Guillemot writes:
Four months after release 1.14, I'm glad to announce the availability of release 2.0. Thanks to everyone who has contributed to this release and particularly to the committers Ahmed Ashour and Daniel Gredler.

From the large list of improvements contained in this release, following are particularly important:

• first release to target Java 5: HtmlUnit now makes intensive usage of Java 5 features, particularly of generics
• HtmlUnit DOM now implements org.w3c.dom making interaction with 3rd party libraries easier
• improved support for incorrect HTML code
• and as always a very large number of improvements in JS support with a major achievement: the GWT 1.4 tests now pass when simulating Firefox as well as when simulating Internet Explorer

This major release contains a few incompatible changes and can't be dropped as a replacement of HtmlUnit-1.14.

Have a look at the changelog for more details.

Changes: http://htmlunit.sourceforge.net/changes-report.html
Download: http://sourceforge.net/project/showfiles.php?group_id=47038

ibiblio.org synchronization request: http://jira.codehaus.org/browse/MAVENUPLOAD-2009

Page on Facebook: http://www.facebook.com/pages/HtmlUnit/10454838218

I had to go back and check the date when reading this article. For a moment, I thought I'd found an old article from the nineties. This sounds just like the anti-Java tirade that we used to hear from the C/C++ people. Only now it's Java people getting upset that they're now the ones being displaced.

I was also amazed that in 2008, people still feel that they have to justify the speed of Java. Yes, it's fast enough.

Two quotes really jumped out at me.

Java also enjoys a 10-times performance advantage over interpreted scripting languages like JavaScript, Perl and Python, he said. "The scripting languages aren't JIT-ted yet," he said.

There are two different things called Java that are being blended together and if we're going to make an apples-to-apples comparison, we have to split them apart. We have the Java language and we have the Java Virtual Machine (JVM). They are very different things and yet we tend to pretend they're the same.

In this case, we're comparing the JVM to the runtimes underneath JavaScript, Perl and Python. This is not a language comparison - it's a virtual machine comparison. Since JavaScript and Python (although not Perl) can all run on the JVM, it's a foolish comparison. Does the JVM use a JIT when it's running Java and then turn it off when running Python code? Clearly not.

All the languages mentioned here can run on multiple different virtual machines. Some of those virtual machines have JITs and some don't. The quote above doesn't seem to make any sense in this context.

Scripting languages are ideal for smaller programs but Java is the choice for larger programs, he said. "As your program grows in size, the lack of strong typing basically kills your ability to handle a very large program and so you don't find the million-line Perl program," he said. One-million-line Java programs are plentiful, Click said.

Didn't we give up on the idea of measuring by lines of code a couple of decades ago? So if it takes a million lines of code to write a given application in Java and it only takes 100,000 lines to write that same application in Ruby or Python or Perl, does it somehow magically make the Java version better because there are more lines?

One million lines of Perl (to use his example) would be a truly massive application. It would have significantly more functionality than the equivalent Java application simply because Perl code is very compact compared to Java code.

Now, I wouldn't choose Perl to create a huge application but that decision has nothing to do with the fact that it's dynamically typed. I would have no hesitation recommending Ruby or Smalltalk for something huge and they're both dynamically typed.

Link from James Robertson

I'll be speaking again at the TSOT Ruby on Rails Project Night. I had a lot of fun at the last one so I offered to speak again

Dynamic finders are one of those things that seem magical to people who are used to statically bound languages like Java or C. In this talk, I'm going to demystify some of this magic so that you can see how one of the more interesting methods in the Rails code actually works.

If you use Facebook, you can RSVP on the event page.

I've been using eclipse with the ruby RDT as my Ruby development platform for quite some time and while the RDT doesn't integrate with eclipse as cleanly as the Java tools do, it's a reasonable environment to work in.

Recently, I was at a talk on Mylyn (which was really impressive) and the speaker mentioned that if you wanted to use Mylyn with ruby projects then you needed to be using the Eclipse Dynamic Languages Toolkit. I hadn't been aware that there was a second plugin for ruby work in eclipse so I made a note to check it out.

I installed the dynamic languages toolkit today and discovered that it has built-in support for both ruby and tcl. It seems to do everything that the RDT was doing for me and the fact that it supports Mylyn is a big win.

I was talking to someone today who was under the impression that ActiveRecord was tied to Rails and couldn't be used unless you were writing a web application. So here's a sample that shows exactly that.

Put this in a file by itself (I called it sample.rb) and run it from the command line with "ruby sample.rb"

require 'rubygems'

require 'activerecord'



ActiveRecord::Base.establish_connection({

  :adapter => 'postgresql', 

  :database => 'mydatabase',

  :username => 'myuserid',

  :password => 'mysecret',

  :host => 'localhost'

})



class Brand < ActiveRecord::Base

end



Brand.find(:all).each {|b| puts b.domain_name}

Obviously, you have to have a database set up and tables defined and all that. The point is that this little bit of code initializes ActiveRecord, creates an ActiveRecord model and starts calling methods on it. All of this from the standard ruby interpreter with no web server in sight.

The parameters that you pass in are the same values that you would have specified in config/database.yml in rails.

So what if I want multiple models? Just define them all inline.

require 'rubygems'

require 'activerecord'



ActiveRecord::Base.establish_connection({

  :adapter => 'postgresql', 

  :database => 'mydatabase',

  :username => 'myuserid',

  :password => 'mysecret',

  :host => 'localhost'

})



class Brand < ActiveRecord::Base

end



class Account < ActiveRecord::Base

end



Brand.find(:all).each {|b| puts b.domain_name}

And what if I want to use all the ActiveRecord magic like associations and validations? Same thing.

require 'rubygems'

require 'activerecord'



ActiveRecord::Base.establish_connection({

  :adapter => 'postgresql', 

  :database => 'mydatabase',

  :username => 'myuserid',

  :password => 'mysecret',

  :host => 'localhost'

})



class Mailbox < ActiveRecord::Base

end



class User < ActiveRecord::Base

  has_many :mailboxes, :dependent => :destroy

  has_many :brands, :through => :mailboxes

end



class BillableItem < ActiveRecord::Base

end



class Account < ActiveRecord::Base

end



class Brand < ActiveRecord::Base

  belongs_to :account



  has_many :mailboxes, :dependent => :destroy

  has_many :users, :through => :mailboxes, :order => 'last_name, first_name'

  has_many :billable_items



  validates_presence_of :domain_name, :business_name

  validates_uniqueness_of :domain_name

end



Brand.find(:all).each {|b| puts b.domain_name}

I will be speaking at this event in a couple of weeks. I'll be showing off some of the innards of EasyBrandingTools.com to illustrate interesting Ruby and Rails stuff. The last event that TSOT put on was a lot of fun so come on out and join us for this one.



Updated: Facebook event page. If you're planning to come and you're on Facebook, please RSVP here.

This interesting announcement came across the rails mailing list this morning.

WSO2 WSF/Ruby team is pleased to announce the release of WSO2 WSF/Ruby 1.0.0.

WSF/Ruby is the Ruby language extension to WSO2 WSF/C. This version enables you to consume/provide Web Services both with REST and with the power of WS-* stack including WS-Reliable Messaging, WS-Security, WS-Addressing and MTOM Attachments.

For more information, please visit our project home page, http://wso2.org/projects/wsf/ruby

What does this really mean?

If this project actually does what it claims (I haven't tried it yet) then it means that Ruby is now a first class citizen in the SOAP based web service world (REST based web services have always worked well with ruby). In the past, if you wanted to work with SOAP based web services and needed WS-Security to provide authentication or encryption, your choices of platforms were either Java™ or Microsoft's .NET. Yes, ruby offered some level of basic SOAP support but higher level protocols like WS-Security just weren't available.

This announcement would imply that those days are past. That if you need to interoperate with a SOAP based web service stack, you can now add ruby as a possible platform. This is extremely good news.

I did some work last year that would have been in ruby had these services been available then.

There is a new blog reading service called Illumio that is extremely badly behaved. Their service is hitting this blog three times a second on average and has used over FOUR GIG of my bandwidth this month pulling RSS feeds.

I'm not amused. If this continues, I'll have to block their ip.

I've sent an email to their support address but perhaps a rant here will make it back to them first. It appears that they starting hitting the blog last month but it's only this month that they've used enough of my bandwidth that I'm starting to get warnings from my ISP.

UPDATE: The CEO of Illumio has personally contacted me to apologize for this and to tell me they're working to fix the issue. This makes me feel quite a bit more positive about their company. I'm still annoyed by the fact that it happened at all but this is clearly a mistake that they're trying to fix. I'll give them points for that.

UPDATE 2: The issue is fixed and I'm impressed by how rapidly and professionally the problem was dealt with.

Phil Windley writes about a bug in Firefoxes implementation of link prefetching. What I find most interesting is that I hadn't been aware that Firefox did anything like this and yet it seems to have supported it for a long time.

From the prefetch FAQ:


Pete Freitag points out that Google uses this technique in their search results to preload the first returned item. So if the result you wanted was the first one in the list, the page will appear to load faster.