It’s been an hour and a half since I’ve started reading the Apache Struts 2 Web Application Development book and I’m going nuts. I’ve dealt with more XML in this hour and a half than most of the time in my degree.

Story told short:

1. I’ve installed Netbeans and an Apache Tomcat 7 server.
2. I’ve created a new Maven Web application.
3. I’ve copied a struts routing example, along with the respective action POJO.
4. Tried to reach the URL; no luck.
5. Looked up for a web.xml example, in order to correctly load Struts.
6. Project won’t deploy. FML.

At an hour and a half, reading and typing, I could have done so much more in Rails or Sinatra. Sorry.

I don’t care what people think about me comparing Java to Ruby at such an early stage in trying to learn. But hey, it’s 2012, software ought to be easier to accomplish. Why does it have to be so difficult to understand how a simple controller is executed? Why is the Java way so convoluted?

I’ll update this post when I can create a Struts action properly, perhaps with an interceptor. If I can make it to apply some TDD along the way, I pat myself in the back. Until then, I stay with the attitude.

This is the first time I’ve ran into something like this and it was interesting to realize what it actually means when developing a business logic in Rails.

Context

In this app I’m building, I have Students that have a Record per year. Each record has several Events. These records also have two specific events: a test and an audition, registered in the schema as id‘s in the record’s table.

So what did technically happen?

I wasn’t able to access those specific events through the associations specified in the model. Given r = Record.first, when I tried to access the audition, by using r.test, Rails would use a SQL query that would correspond to r.events.first instead.

After acknowledging that, I turned to @varandas and we both thought it might be a bug in the Rails framework. Turns out it wasn’t; all I had to do was switch from has_one to belongs_to (thanks @drogus!). The reason for that is the foreign key is on the records table. From the framework’s perspective, it looks like the record actually belongs to the event, when in practice it’s not.

Code sample

TL;DR – Being bossed around compromises inovation, creativity and it destroys one’s true sense of purpose. Don’t let that destroy you.

The Piscean Era has begun around 0-400 A.D and, according to the math, will last until very soon. The Pisces sign has several connotations: sacrifice, martyrization, faith, dogma and spiritual dictatorship. This is still a challenge for many creative and passionate developers who wish to spread wings and bring a better world through technology. Only the bravest survive the food chain and those won’t don’t become slaves.

The Aquarian Era follows the Piscean. It stands for freedom, love and conciousness. These are the values I fight for. My goal is to build great stuff on the web, the way I feel it should be done. A lot of what’s built today looks great, but there are still major issues to tackle when it comes to fulfilling needs. Those needs are overrated and they don’t help the smallest, the poorest and the less bright; they only serve the allegedly powerful and certainly wealthy.

Open source is an awesome Aquarian movement. It’s not just the code, it’s a whole mindset. It lets people make their move and be rewarded for it. That reward is not just money, it’s also the joy of the contribution and hope for evolution.

This is not about companies anymore, each team member is too important to be masked with a company logo. People move a lot these days, there’s actually no point in fixating all the credit in one company in a point in time. Letting developers choose what they want to do is crucial to creativity. They will do what it takes if they feel it’s the right thing to do.

Context

I play the Interaction Designer role and digging deeper to Struts and Java servlets, more than just HTML/CSS/jQuery, is becoming quite the challenge. I happen to love Rails and would love to try it out. My main goals are:

1. Take advantage of Rails’ controllers, routes and views;
2. Use Haml with Markdown and Sass with Compass;
3. Take advantage of REST.
4. Finally use Rspec and inject some proper testing and specification practices.

All of this while the rest of the team can still code the models (aka the business logic) the way they prefer. I think it’s a good cause and it’s a great opportunity to improve development happiness and systems integration.

The challenge

Here is a list of the things I know:

1. I can use RVM to quickly install and use JRuby.
2. Gems with C extensions are a bit difficult to use inside the JVM.
3. There are alternative gems to the ones above.

I also need to get in touch with these concepts:

• Web application development in Java: .jar’s, war’s, etc.
• Tomcat or Glassfish as servers.
• Importing Java code into a Rails or Sinatra app.
• (optional) Using Ruby code from Java.

I guess the main question would be: Where do I start? What am I supposed to know to begin developing some sort of Rails app, using it under Tomcat or Glassfish and reusing Java code from the rest of the team? Thanks for helping me, guys!

A couple of weeks ago I needed to build a plugin for DT: MultipleFilters, something that would allow me to filter data by a range. Each column would have a pair of fields and I would have to mix them into the bowl of parameters to send to the server . It looked fun, so I started doing it, hoping to achieve a beautiful Javascript prototype object that would do magic when plugged in by a single letter. As I was finishing building it, I realized some issues had to be fixed.

Disclaimer: The plugin does not approach local processing, only server-side.

Push data to the server without using fnServerData

With minimum coupling in mind, I wanted to find a way that allowed to push data to the server without using the fnServerData option in the initializer. That’s not what the end user wants, he wants to just download the plugin, place it in and there, it does range filtering.

My suggestion here would be an array of functions (perhaps called afnbeforeSend) that allowed any plugin builder to be able to inject parameters to be sent. That way, when my plugin is initialized, I can push my function that injects such parameters and I don’t pollute the initializers.

Integrate with other plugins, such as ColVis or ColReorder

Datatables’ features that have controls, such as MF, require a DOM object to be returned in the fnInit function. As of now, I am generating the feature with just a button. The actual controls are being appended to the table’s footer, with no regard to the column order or state of visualization.

At some point in time, I faced a lot of trouble as I watched multiple controls get spawned across the table, as well as the controls not being switched along with the column being reordered nor being hidden / shown when you check a column with ColVis. Wouldn’t it be great if both ColVis and ColReorder had some sort of array of callbacks we could populate? When ColVis toggled a column, I could access it and toggle my filter controls accordingly. When ColReorder dragged a column around, I could know which one and to where it moved. All of this without requiring additional initialization and granting cohesion, based on a functional approaching.

For what it’s worth, I actually had to hack inside ColReorder and add some code in there so I had to accomplish some integration between MF and CR.

So, is there a need for better integration?

Before I jump into the conclusion, I should say that Allan did a great job on building Datatables, it’s a remarkable piece of software.

I told him that I would risk being stupid to the point of being told to RTFM, but I stand for ease of read as well as ease of use. There should clearly be some plugin development guidelines and I personally volunteer to help build those guidelines. Based on what I know, I would say there is a need to improve integration. If it’s not for how it’s built, it would by the way it’s told.

I told her I was looking forward to studying Cucumber and Rspec in order to improve my skills. She asked me how I did documentation using those tools and I answered: “Cucumber and Rspec are the documentation”. As she twisted her nose I could tell I had to research more on the topic; so I did.

I have been reading the Rspec book. Definitely read it if you want to know how to code better.

Why does “documentation” feel so hard?

Dumb answer: It sucks. Smart answer would be something like: You might be doing it wrong.

Documentation specifies a software’s collection of procedures it performs. Why should we even try and build software that does not resemble people’s expectations? (If it weren’t Ruby, I’d guess the software to be built required another form of understanding, which is fine and yet irrelevant in this case.)

Who is the documentation for?

Not just people, but stakeholders who want straightforward answers. 150 pages of paper is not the way to accomplish that, IMO. They know what they asked for and we should build the answers accordingly.

Cucumber and Rspec

What’s wrong with automated documentation? How beautiful is it to have a stakeholder look to a set of Cucumber features and Rspec examples that assert precisely what he asked for — damn, he might have written those features…! — and find the answer?

Cucumber defines vocabulary for the system, states what it does and who. And it helps developers and stakeholders connect and collaborate.

PS: being a designer, I know I must present my documentation as pleasant as possible. Well, I guess Cucumber and Rspec just support this statement of mine, don’t they…?

I have reached a milestone. The next thing I need to build is an index page that inevitably fetches all the posts I have. But how am I supposed to do that with files? If I didn’t build Blogrite to allow multiple content providers, I would fix the posts’ location into the filesystem and perform system calls like find or tail or something else. I wouldn’t have to think about an index file that would be built everytime I write something new, in order to increase performance. But since this is not the case, I need to come up with a solution.

What I do know is the responsability of getting the index content is the provider’s.

• If I hosted content live along with the system – like I mentioned above –, find or tail system calls would be fine.
• If I used Git, I could use a post-commit hook and have it change that post info in an index file.
• Dropbox would be a totally different story. Are there any hooks in the service for automated scripts? (honestly I don’t think there are, feel free to prove me wrong!)

All of this considering an automated index page. There might be another ways of accomplishing this. I’m open to suggestions, throw ‘em away.

I asked Tolga Arican what the license what kind of license they had for the plugin, they said none. I decided to fork the project and host it on Github, so I could make improvements to the plugin and share them with everyone. Enjoy! Props to the Utopic Farm team for building such a great plugin.

I told you before I would post a video on the development of Blogrite after Rubynorte. So here it is.

Topics included

• It works!
• Well, sort of. The first request is a 200, all the other ones throw me a 404 CSS file!
• ERB for rendering the templates.
• Classes over modules for extension?

I appreciate your feedback, you can clone the project at Github.

After fixing the bugs I still have, I intend on fetching the list of posts at the root. Or I can let the final user create an index.txt for a landing page.

I have talked a little faster today but still a bit longer than I had wished for. I’m excited to bring good news about Blogrite, I’m steadly growing my knowledge on Ruby. Rack is really interesting, I’m loving how simple and easy to learn it is so far.

Testing is fun when you start to think of features you want to implement and quickly draw a quick test that will fail and you try to make it pass. I can’t wait to start the refactoring phase, when everything is working the not-so-right way.

Thanks for all the help guys.

What I noticed already

• I chose Minitest after Rspec because it’s leaner, which is what I want for now.
• Requiring ‘minitest/autorun’ will run my tests automatically.
• The class to inherit from is Minitest::Unit::TestCase, not from the Test module.

I’m still missing stuff like how to require and autoload right. I’m pretty sure is wrong with that.

@dhh started a heated discussion on Rspec vs Test::Unit, right after I finished the recording. I must admit he’s a genius and he actually made me reconsider Test::Unit again, since it is considerably lighter than Rspec. I’ll have to think about it.

• PHP is not it anymore.
• WordPress or Drupal are top notch. I don’t need top notch; I want to publish content easier and faster.
• I want to learn Ruby the hard way. I want to learn how to test, I want to learn how to build a gem, how to use Rack, all of it. Only then I’ll be more confident enjoying the full power of Ruby.
• I don’t want to depend on a database to publish my content. I never saw that much sense in storing articles inside a relational database column. I love Git and it seems it is capable of storing the content I want.
• I’m sick of 10+ spam comments a day.

My next dare: a git + ruby based blog. So, I’ll be posting several articles on how I’m doing. If possible, I’ll post videos too. I’ll make sure to use Vimeo since it accepts OGG (it seems lighter).

Thanks to the people at the Ruby meetup in March 19th who have motivated me (even without knowing).

If you’re a Rails developer / designer, why even bother? Why not just start pumping the business up? You don’t have that much attrition to get an app up and running. All the time you spend in prototyping should be invested in building the actual app.

Web application design walks with its development, they are too close to be set apart.