edovale's blog

API design.

2008-10-16T20:46:00.000-07:00

I find this entry very useful. So much so that it is worth printing and sticking somewhere it can be read while coding APIs.

Evil Ike

2008-09-21T11:06:00.000-07:00

If this is how Ike, being a category 2 hurricane, left Texas; I can't think if the disaster it left across Cuba where it hit as a category 3 hurricane. Very impressive pictures of Texas though.

Hibernate & maven.

2008-09-15T20:06:00.001-07:00

Some time ago a blogged about some annoyances in the, at the time, current pom.xml for hibernate. At that time the maven metadata for hibernate was maintained by the maven community who could not keep up with the amount of work generated by updating dependencies to hundreds of software libraries.
Well, things have changed quite a bit since then. 2 of the most important java open source projects of all times have started maintaining and publishing their own maven artifacts!! Spring Framework announced it here on January/2008 and more recently Hibernate did it here.
It is clear that maven is gaining ground in the build market when projects of this caliber start supporting it. In the case of Hibernate they totally migrated their build from ant to maven.
I am glad that maven has gotten to this point.

Consistency vs Improvement

2008-08-21T19:55:00.000-07:00

On every medium to large software system, consistency becomes critical.
By consistency I mean the application of certain architectural patterns that are considered the standard way of "doing things" in a software organization. Consistency helps keep the system understandable to all programmers currently involved, and to those that may come in the future. Consistency also helps when solving known problems as established patterns can be applied with ease. Consistency attempts to keep the entropy of the system under control.
On the other hand, it can also make improving a system hard. How can you change the standard without introducing inconsistencies? In a large project there is probably no good reason to change existing code that is already working for new code that does the same in a different way. No matter how better the new approach is; it just isn't cost effective.
In my opinion, consistency can not be taken in an absolute way. Consistency is never a good reason not to improve a design. If this wasn't true there would be no legacy code ever. Instead new valuable ideas should be embraced and considered the new consistent way. No new code should be written following the old way. Changing the old code should be considered technical debt if it merits the effort which very likely it will. Resistance to changes that improve the quality of a system overall is irresponsible.
How is this issue dealt with in your organization? Is consistency used as a reason to reject changes?

YubNob is kinda cool

2008-06-10T09:35:00.000-07:00

This is a nice idea for shell like search commands from within firefox.

About SpringSource's marketing

2008-05-29T19:47:00.001-07:00

Look what i just got today:


from Kylie Clement 
to me
date Thu, May 29, 2008 at 12:19 AM
subject SpringSource Application Platform
 
Erick,

Thank you for your interest in SpringSource Application Platform, an excellent alternative to J2EE application servers.

Are you available Thursday or Friday of this week for a telephone conversation?  I would like to understand your requirements in detail
and answer any questions you may have.

Regards,
Kylie

Kylie Clement
Regional Sales Manager
SpringSource, Inc.
1875 South Grant Drive, Suite 650
San Mateo, CA 94402

This happened after I registered for a webcast about the newly released Spring Application Platform. I have given my information to the big boys before (IBM, Oracle, etc etc) and never got an email as intrusive as this before.
Anyway, being consistent with the inversion of control pattern, my answer is: No Mr Clement, do not call me; I'll call you if I ever need your help. And no, you will not be able to understand my requirements in a phone call as not even I am sure of what they are.

maven-js-plugin

2008-05-22T21:18:00.000-07:00

Wao,
I was blown away by the ease of use of this maven plugin.
Just add it to your pom.xml lie this:

<plugin>
  <groupId>com.mobilvox.ossi.mojo</groupId>

  <artifactId>maven-js-plugin</artifactId>
  <version>1.3.1</version>
  <configuration>
    <mergeWarFiles>true</mergeWarFiles>

    <classifier>js-compressed</classifier>
  </configuration>
</plugin>

And then in the maven output, during the package phase, you get to see something like this:


[info] JavaScript file compression complete.
[info] Size of JavaScript files after compression: 1612632 bytes
[info] Compression ratio: 27.103024264908303%
[info] Creating WAR: mywar.war with compressed JavaScript

So there you go, another reason to love maven.

Maven, love it, hate it, love it, hate it, ......

2008-05-22T13:49:00.000-07:00

Lately I've been doing lots of maven stuff.
I just released a new version of the web app I am working on and this version is going to production. Releases always are maven intensive tasks (for maven based build systems obviously), package the app, make sure data population scripts and schema creation ones are all good, deploy to new environment, run tests against it to ensure everything is ok, etc etc.
Tweaking maven to behave appropriately can sometimes be frustrating although, in this opportunity, I have to say that almost everything went smoothly.
One thing that I find tricky is the build profiles. I am obsessive with DRYing every single line of code i write and profiles do not make it easy in this regard. A feature I would love for maven to have is the ability to activate a profile from within another profile.
For example, I want to have a profile called mysql that defines the dependency on mysql, plus properties such as driver class name, hibernate dialect class name etc. I want to have another profile to define an environment, say nigthly as an example. Nightly consists of a server running an instance of tomcat to which I want maven to deploy the app every night and, a mysql server instance which will host the app's DB.
I don't wanna say mvn deploy -P nightly,mysql; instead I just wanna say mvn deploy -P nightly and the nightly knows to activate the mysql profile.
Overall I like maven and it is a very lively project with a very active comunity. I also like the fact that it has many detractors as this works in favor of making it better. I do wish it was released more often though.

Wolfenstein on JS!!

2008-04-30T07:06:00.000-07:00

Outstanding!!!!
Things like this tell more about the power of javascript than all the books on the language out there.

OS X Focus-follows-mouse issue

2008-04-27T10:02:00.000-07:00

Sometime a go I bloged about this issue. Now Stevey has a very detailed blog entry about this issue

Validation Framework for java

2008-03-25T13:47:00.000-07:00

Wouldn't it be nice to be able to do something like this:

public class Address {

   @NotEmpty @Max(50)
   private String street1;

   @Max(50)
   private String street2;

   @Max(9) @NotNull
   private String zipcode;

...

}

and then be able to say something like:

public class Order {
   @Valid @NotNull private Address delivery;
}

The implications of something like this go beyond mere validation. IMHO it si an elegant solution that if built properly will improve the DRYness of the code across all layers of any java application.

This is the intention behind jsr-303. Emmanuel Bernard from jboss has more details here.

Another Mac OS annoyance

2008-03-17T10:13:00.000-07:00

For some reason in Mac OS, if an application does not have the focus, you would have to click twice on it to get something done. The first click will give the application the focus and the second will do what you where intending to do in the first click. In windows this is not true so for someone coming from Windows, like me, this is particularly annoying. Nonetheless it is not the end of the world.
Mac OS is definitely faster than windows for the kind of work I do on it. I am just finding it a little bit tricky to get use to its windowing system.
It could be the case that, for someone with little or no computer experience; getting use to this kind of things would be easier than it is for someone coming from a different OS.

Amigo camino a la fama.

2008-03-06T19:35:00.000-08:00

Tengo un amigo cineasta de hace mas de 20 anos que, con su opera prima como director, parece haberle acertado al clavo en la cabeza.
La verdad yo siempre crei en el. Muchos no, y no los culpo por que despues de empezar y dejar inconclusas carreras de musica, psicologia y filologia, hay que tener mucha fe en alguien para seguir creyendo que podria llegar a algo. Finalmente encontro se verdadera profesion como cineasta lo cual me hace extremedamante feliz.
Ale, te deseo toda la suerte del mundo, aunque se que, como hasta ahora, la suerte te la haras tu mismo.

Software Market works upside-down

2008-03-06T15:11:00.000-08:00

What an interesting take on the software market this is.

No mnemonics in MAC OS X?

2008-03-03T11:50:00.000-08:00

I just got a new mac book pro!! I am extremely excited about it.
I have always been a windows guy; and not, I am not exactly ashamed of it. As a developer, the majority of my career has been working on the java platform even though windows was my platform of choice for development.
Anyway, now I find my self lost around mac os x. I have to say that I am having a blast learning to work with it but today I think I found something that in my opinion is pretty serious.
As I started coding away on eclipse in my new mac I went to create a new test class and, to my surprise, eclipse has no mnemonics at all on the mac os platform. On windows I'll go shift+alt+n+j+j+enter to open the new test class dialog, alt+o to select the source folder and alt + f to finalize the wizard. On the mac I'll have to go back to using the mouse for this and many many other things I do just with the keyboard in windows.
At the beginning I thought I was missing some hack in the eclipse configuration to activate this but then realized that iTunes, or any other native mac application for that matter, have no mnemonic support at all. Then I called a fellow developer who is a big mac advocator and he pointed me out to this
(search for mnemonics) and here clearly states that mnemonics are not inline with Aqua guidelines.
Am I the only one who makes heavy use of the keyboard? How come no mac fan ever complained about not been able to drive his/her IDE from the keyboard??

Agile without Tests???

2007-11-22T12:24:00.000-08:00

Can a software organization where the majority of developers have no idea about TDD be considered agile?
I think no, but I may be wrong. What do people out there have to say about this?

darn hibernate pom.xml

2007-05-11T22:10:00.000-07:00

I just finished figuring out a problem for which I could not find any explanation online. It all started happening when I decided to take spring2 aop for a whirl. The app I am working on is a web app using spring and hibernate; both of them to the latest versions which 3.2.x for hibernate and 2.0.4 for spring.
My build system is maven based. The things I love about maven (for regular developer tasks)is its dependency resolution capablities as well as the eclipse:eclipse plugin which helps me setup eclipse in no time.
When I added a dependency on spring-aop and spring-aspects my application stopped deplying complainig about a class in asm that was not found. The exception was comming from hibernate trying to proxy an object using cglib. This was the stacktrace I was getting:
java.lang.NoClassDefFoundError: org/objectweb/asm/CodeVisitor
at net.sf.cglib.core.KeyFactory$Generator.generateClass(KeyFactory.java:165)
at net.sf.cglib.core.DefaultGeneratorStrategy.generate(DefaultGeneratorStrategy.java:25)
at net.sf.cglib.core.AbstractClassGenerator.create(AbstractClassGenerator.java:216)
at net.sf.cglib.core.KeyFactory$Generator.create(KeyFactory.java:145)
at net.sf.cglib.core.KeyFactory.create(KeyFactory.java:117)
at net.sf.cglib.core.KeyFactory.create(KeyFactory.java:108)
at net.sf.cglib.core.KeyFactory.create(KeyFactory.java:104)
at net.sf.cglib.proxy.Enhancer.(Enhancer.java:69)
at org.hibernate.proxy.pojo.cglib.CGLIBLazyInitializer.getProxyFactory(CGLIBLazyInitializer.java:107)
at org.hibernate.proxy.pojo.cglib.CGLIBProxyFactory.postInstantiate(CGLIBProxyFactory.java:43)
at org.hibernate.tuple.entity.PojoEntityTuplizer.buildProxyFactory(PojoEntityTuplizer.java:162)
at org.hibernate.tuple.entity.AbstractEntityTuplizer.(AbstractEntityTuplizer.java:135)
at org.hibernate.tuple.entity.PojoEntityTuplizer.(PojoEntityTuplizer.java:55)
at org.hibernate.tuple.entity.EntityEntityModeToTuplizerMapping.(EntityEntityModeToTuplizerMapping.java:56)
at org.hibernate.tuple.entity.EntityMetamodel.(EntityMetamodel.java:295)
at org.hibernate.persister.entity.AbstractEntityPersister.(AbstractEntityPersister.java:434)
at org.hibernate.persister.entity.SingleTableEntityPersister.(SingleTableEntityPersister.java:109)
at org.hibernate.persister.PersisterFactory.createClassPersister(PersisterFactory.java:55)
at org.hibernate.impl.SessionFactoryImpl.(SessionFactoryImpl.java:226)
at org.hibernate.cfg.Configuration.buildSessionFactory(Configuration.java:1294)
at org.springframework.orm.hibernate3.LocalSessionFactoryBean.newSessionFactory(LocalSessionFactoryBean.java:805)
at org.springframework.orm.hibernate3.LocalSessionFactoryBean.buildSessionFactory(LocalSessionFactoryBean.java:745)
at org.springframework.orm.hibernate3.AbstractSessionFactoryBean.afterPropertiesSet(AbstractSessionFactoryBean.java:134)
at org.springframework.beans.factory.support.AbstractAutowireCapableBeanFactory.invokeInitMethods(AbstractAutowireCapableBeanFactory.java:1175)

and more stuff which adds no value to this blog entry.

The problem, after hours of digging, banging, pulling hair and snapping was the hibernate's pom.xml has cglig-2.1_3.jar as dependency and somehow this guy depends on asm-1.5.x.jar. Spring aop depends indirectly of asm-2.2.3 and maven, smart enough, was using the latter is its a newer version. To make a long story short, I added an exclussion to hibernate dependency for cglib and added cglib-nodep as one of my dependencies and this solved the problem.
I hope, if anyone hits this problem ever, he/she will google the stacktrace and will find this post.

Validation Framework For Commands.

2006-11-01T00:50:00.001-08:00

In my company they came up with a rather strange way to"customize/extend" the product. There is a sort of framework upon whichcustomizations are built. The way to customize a certain piece of logicis by using a variation of the chain of responsibility pattern. Pieces of logic are put into command objects and these objects are chained using Apache Commons-Chain. The reasons why this done like this is a topic for another post.
Becausethe application is spring based, it was awkward to have these commandobject instantiated outside spring. By doing this we were unable toleverage the dependency injection facility in spring and therefore, hadto come up with a very wacky way of doing this which basically wasputting collaborators into the command context.
Recently this wasfixed. We are now able to declare command objects as spring beans andinject anything we want into them. The same goes with chains. This isindeed, a great improvement over what we had before.
Now, to thepoint of this post. In my team, we thought it would be nice to have aframework to validate the command contexts before the actual executemethod is called. By having this, we'll be saving our selves theclutter of having validation code inside the commands; instead, thecommands would assume everything expected to be in the context wasthere, and focus only on getting the job done.
This is the kind ofuse cases aspects are very well suited for. It is a cross cuttingconcern because validation has to be performed across a large set ofobjects.
So, we came up with a before advice for the execute methodon the command objects and also wrote a factory that is able to returna validator object for a particular type of command. This factory onlyhave to be declared as a bean in the spring context and wheninitialized it will search the application context for Validator beansand put them in a map keyed by the type of comand they know how tovalidate.
If the factory returns nullfor a particular type of command then the advice will let theinvocation continue without further actions. At the most, it logs awarning about the missing validator.

This is how the advice looks in real life:

public class CommandContextValidationAdvice implements MethodBeforeAdvice {
private CommandValidatorFactory validatorFactory = null;

/**
* Lookup a CommandContextValidator via the factory. And if not null then execute the validate method.
*/
public void before(Method method, Object[] params, Object command) throws Throwable {
CommandContextValidator validator=validatorFactory.getCommandValidator(command.getClass());
if(validator != null) {
validator.validate((Context) params[0]);
}
}

/**
* @param validatorFactory The validatorFactory to set.
*/
public void setValidatorFactory(CommandValidatorFactory validatorFactory) {
this.validatorFactory = validatorFactory;
}
}

The second problem we faced was applying the aspect to the commands. We didnot wanted to change every single command bean definition as this woulddefeat the purpose of making the validation framework orthogonal to theactual command hierarchy.
In order to solve this, we used the autoproxy facility in spring. The bean definitions in the applicationcontext then looks like this:

<BEAN class=org.springframework.aop.framework.autoproxy.BeanNameAutoProxyCreator id=jdkBeanNameProxyCreator>
     <property name="beanNames">
          <VALUE>*Command</VALUE>
     </property>
     <property name="interceptorNames">
          <LIST>
               <VALUE>commandContextValidationAdvisor</VALUE>
          </LIST>
     </property>
</BEAN>

<BEAN class=com.mycompany.myclientcompany.command.CommandContextValidationAdvice id=commandContextValidationAdvice>
     <property name="validatorFactory">
          <REF bean="commandValidatorFactory"/>
     </property>
</BEAN>

<BEAN class=org.springframework.aop.support.RegexpMethodPointcutAdvisor id=commandContextValidationAdvisor>
     <property name="advice">
          <REF local="commandContextValidationAdvice"/>
     </property>
     <property name="pattern">
          <VALUE>org.apache.commons.chain.Command.execute</VALUE>
     </property>
</BEAN>

The first bean definition is the auto proxy creator. It will proxy everybean in the application context whose name end with the word Command. In our case this was not much of a problem because all the commands were following this convention even before we thought about the validation framework. The rest of the bean definitions are the Advice and the Advisor respectively.