Raible Designs

Introduction to Grails with Scott Davis at the Colorado Software Summit

According to Scott, today's Java-based web frameworks are partial solutions at best. They're only solving one piece of the puzzle - you still need to manage persistence, deployment, etc. all by yourself.

We're moving into a new era of web frameworks. The expectation now is a full-stack solution. Grails is a fully integrated modern Java web application in a box. It contains Spring, Hibernate, Quartz, Log4J, Jetty, HSQL, JUnit and Ant. You're not limited to using Jetty, you can type "grails war" and create a WAR that you can deploy to any application server. In a single zip/tar, you get the whole thing - including the database and servlet container. You get a lot of good default for free, but you're not limited to those defaults.

What does "modern" in a framework mean? It means it uses Convention over Configuration - Grails autowires components together based on naming conventions. Struts 1.x uses Configuration over Configuration - the more XML the better.

The Wisdom of Crowds - why are the many smarter than the few? What is the wisdom of crowds when it comes to web frameworks. Struts is the wisdom of crowds when it comes to web frameworks. It's based on sound principles (MVC) and was written by Craig McClanahan. He was the architect of "Catalina" Tomcat 4 and wrote Struts shortly thereafter. David Geary was contributor #2 to Struts. It has a proven track record and has a 60%-70% market share. Struts must be the perfect framework - especially since it has such a great pedigree.

So what's wrong with Struts?

• It's 7 years old
• It's verbose and overly complex (reflective of the EJB 2.x era)
• Splintered community: Craig moved on to JSF -> created Shale, then the WebWork merger
• While Struts 1.x was an unqualified success, Struts 2.x can't seem to build that critical mass - it can't seem to reach The Tipping Point like Struts 1.x did

The Recipe for a Tipping Point: Contagiousness / viral, tiny "tweaks" to a proven model can yield big, disproportionate effects, the rise is not "slow and steady" - the effect is dramatic and immediate.

Ruby on Rails won the hype award. It's largely worth the hype, but it's not revolutionary - it's evolutionary. It has tiny tweaks to the proven MVC-driven approach: Convention over Configuration, Scaffolding and Unified Technology Stack.

The one thing that Rails is lacking is Java support. Grails, on the other hand, offers the same experience using known, proven Java solutions.

• Rails => "replacement"
• Grails => "upgrade"

Scott drank the Rails Kool-Aid for a while and enjoyed it, but found it difficult to switch from Ruby in the morning to Java in the afternoon. With Grails, he doesn't have to do as much context switching, as well as all the Java libraries are available - the ones you know and love.

You can drop groovy.jar into your classpath and write Groovy code. One nice thing about Groovy is you can rename your existing .java files to .groovy and they'll work just fine. Included Ajax support: Script.aculo.us and Protoype as well as YUI. YUI is battle-tested since it's used by Yahoo and very well documented. You can use "grails install-dojo" to install the Dojo toolkit. Grails has a wealth of plugins available at http://grails.org/plugins.

Now we're going to crack our nuckles and build some code - slides are over.

 grails create-app conference
 cd conference
 grails create-domain-class Speaker (add some fields)
 grails generate-all Speaker
 grails run-app

The impressive things about this set of commands is I was able to 1) download Grails and 2) run all these commands at the same time that Scott did. I was definitely impressed (I knew I would be). Auto-scaffolding - you can get the same thing as "generate-all", but it generates controllers and views at runtime in-memory. All you need to do is create a domain object (i.e. Talk) and then create a TalkController that has the following line in it:

def scaffold = Talk

As a test, I tried this at the same time that Scott did and got the following error. It looks like Grails/Jetty isn't smart enough to pick up new classes as they're added.

[263102] commons.DefaultGrailsApplication Class not found attempting to load class Talk
java.lang.ClassNotFoundException: Talk
        at java.net.URLClassLoader$1.run(URLClassLoader.java:200)

Grails comes with a number of environments. Dev (the default) auto-reloads changes to Controllers, Views and even the Model (this is helpful for rapid development). Prod loads all items statically for maximum performance. To change the environments - you can change DataSource.groovy.

For some reason, adding/removing some properties on my Talk object and my application hasn't been the same since. Now when I try to access my TalkController, I see the following stack trace:

org.springframework.orm.hibernate3.HibernateQueryException: could not resolve property: difficulty of: Talk; nested exception is org.hibernate.QueryException: could not resolve property: difficulty of: Talk
	at org.springframework.orm.hibernate3.SessionFactoryUtils.convertHibernateAccessException(SessionFactoryUtils.java:640)
	at org.springframework.orm.hibernate3.HibernateAccessor.convertHibernateAccessException(HibernateAccessor.java:412)
	at org.springframework.orm.hibernate3.HibernateTemplate.execute(HibernateTemplate.java:378)
	at org.springframework.orm.hibernate3.HibernateTemplate.executeFind(HibernateTemplate.java:342)
	at org.codehaus.groovy.grails.scaffolding.DefaultScaffoldDomain.list(DefaultScaffoldDomain.java:112)

I asked Scott about this error and he proved that removing properties from domain objects should work. I told him restarting Jetty didn't fix the problem and he suggested dropping the "Talk" table and letting Grails re-create it. Unfortunately, I have no idea where this database is, so that's difficult to do. Doh - now I realize what was causing the problem. Before I dropped the "difficulty" property, I had clicked on the column and that property was still referenced in the URL. When I'd refresh the browser, the stack trace occurred. I don't know if I'd consider this a bug or not.

dbCreate = hibernate.hbm2ddl.auto is used. When in development create-drop is used. In production and test environments, it uses update and saves the data between restarts.

You can turn off Hibernate's automatic schema alteration by commenting out "dbCreate" line in DataSource.groovy.

To run your application in Tomcat instead of Jetty, you can run "grails war" and copy the WAR to Tomcat. The WAR is created with the production environment by default, so you may need to pass in arguments or set environment variables if you want the WAR created in dev mode.

At this point, my battery died. Scott continued to cover how to order form elements (when using scaffolding) with "static constraints" and how to add validation rules. It was an excellent presentation and Grails definitely looks like a really cool web framework. The best part is I learned most of what you need to know to use it - in an hour!

I might have to try Grails soon - I love the concept of Life above the Service Tier and Grails would work nicely for serving up REST. I think YUI, GWT-Ext and Flex are probably the best frameworks for developing a SOFEA client. The question is - when using YUI, how do you download all pages in the application at once?

RE: Trolling with Java Web Frameworks

Fred Daoud has an excellent post titled Trolling with Java Web Frameworks. Well written Fred!

Google Goods at the Colorado Software Summit

This morning, I woke up early and attended Dion Almaer's talk on Google Gears. Dion works at Google in the Developer API group and is a member of the Google Gears development team. This presentation is called How to take your Web Application offline with Google Gears.

Dion starts with a video that Google Developers made. It's a parody of Dick in a Box, but it's called API in a Box. This was by far my favorite part of the presentation and it all went downhill from there.

Gears is a browser plugin that adds a database and a JavaScript API that allows applications to "go offline" and use these resources to store data. It does not do anything to handle synchronization of data back to the online database.

Gears has three modules as part of its API: LocalServer, Database and Workerpool.

The Database is embedded using SQLite. Google contributed Full Text Search to it and the entire database is 250K. Below is an example of how you might use the API:

var db = google.gears.factory.create("beta.database", "1.0");
db.open("database-demo");
db.execute("create table if not exists ...");
...
var rs = db.execute("select * from Demo order by Timestamp desc");
while (rs.isValidRolw()) {
    var name = fs.field(|);
    ...
}

During the rest of the talk, I tuned in and out, but caught a number of interesting tidbits:

• GearShift - DB Migrations for Gears.
• Database Query Tool - one-page browser-based application that can be added to existing applications to add a query browser.
• Gears in Motion - full database tool for creating databases, modifying tables, inline-editing, etc. After building your database, you can export the SQL for use in a Gears application.
• Syncing is not a part of Gears because it's a very difficult problem to solve - especially when you have huge data.
• The best way to implement synching is to start simple, like Zoho Writer. It's like Google Docs, but allows you to read your documents off-line. The next version they're planning on allowing offline editing.
• You should plan on allowing your application to run - possibly by using cookies to store the data and create a "CookieBaseContent" implementation that gets chosen instead of "GearsBaseContent". In other words:

  content = hasGears() ? new GearsBaseContent() : new CookieBaseContent();
  

• Debugging is a pain - you can simplify by using helper functions that allow you to clear the database.
• Google's Read/Write JavaScript API looks like some really cool stuff. It solves the cross-domain problem and is (at its core) a clever bundling of browser hacks. It's not specific to Google's APIs and can be used in any Ajax application.
• Gearsmonkey - uses Google Gears with the Firefox Greasemonkey plug-in to take other's applications offline.
• Dion uses Greasemonkey to add keystrokes to GMail and remove ads from Facebook.
• Wikipedia has an offline version that's powered by Gears. It uses iframes to cache pages with Gears and was developed by Google. It's unannounced by Dion didn't say I couldn't blog about it.
• Dojo offline has a sync framework.
• Vortex is an offline toolkit that builds on Gears.
• GWT has Gears support - all you have to do is drop a JAR in your classpath. If you don't like using Gear's JDBC-esque API - maybe you can use Hibernate with Gears?

Dion ended by talking about how Adobe Air is great for desktop-like applications that you can easily build with Ajax technologies. Silverlight is impressive, but only for media applications - you have to draw components yourself. Java Applets may make a comeback. The browser plugin has been rewritten to be fast as well as have full support for Java Web Start. It's possible that Gears + the Java Plugin can make it possible to use Java technologies (i.e. Hibernate or JPA) to talk to the browser's database. Firefox and WebKit are adding database components to their next major releases - so offline applications should become even easier to develop in the future.

Overall, this was a great talk - largely because Dion is a great speaker and made it fun and interesting.

After Dion's talk, I delivered my Web Framework talk and had some lunch while trying to get Rockies tickets (no luck). After lunch, I attended Bill Dudney's Comparing Spring & Guice talk. I learned some things about Guice I didn't know and enjoyed his comparison of the two Dependency Injection frameworks.

One question that Bill couldn't answer is how Spring 2.5's annotation support stacks up against Guice. Is it as full-featured as Guice? Does it add additional features and keep up with Guice for performance? What about wiring up objects without annotations - does it allow you to autowire your classes based on naming patterns without annotations in your code? What I'm hoping for is a DI framework that allows me to autowire classes using rules/conventions rather than annotations. I'm fine with using annotations for edge-cases, but it seems like a lot of the DI I do these days could be configured up-front and used for the entire application (rather than having to wire up each class).

Overall, it was a great day at the Colorado Software Summit.

Choosing a JVM Web Framework

This morning I delivered my Choosing a JVM Web Framework presentation at the Colorado Software Summit. One of my goals for the talk was to get more audience participation and stories of how folks thought it best to choose a web framework. The room was packed and the crowd was interested, so we barely even used the slides I prepared. One of the most interesting things about the audience was that over 60% (of 50-60 people) were using Struts 1. Most of them came to learn about frameworks they might think about migrating too. Unfortunately, I didn't talk a whole lot about the frameworks and their features, but a few members had advice concerning frameworks *not* to use based on their experiences. Overall, it was a lot of fun to interact so much with the audience and hear their thoughts on web development for the Java Platform.

You can download a PDF version of my presentation from my presentations page. Thanks to all the folks who responded to my Stories Wanted post - I used many of these comments as part of the "Case Studies" in the presentation.

Interface21 on Open Source

Rod Johnson in Replies to Nonsense about Open Source says that Interface21 is the only legitimate company that can offer support for Spring.

...at least that's my interpretation...

Ben Speakmon (of SourceLabs) responds with Nonsense about Interface21.

Both articles are good reads. However, I think Ben has a good point:

One final point for Rod: why did you open source Spring at all? If you're so convinced that no one else can offer credible support for it, why not just make it proprietary?

Is Interface21 becoming the JBoss from two years ago? Will they one day make it difficult for companies to provide services around Spring like JBoss has? Fleury and Johnson will say that "professional open source" is the only way to have a truly successful project. While it may be working well for them, I tend to like DHH's stance on Rails a bit more:

I believe a Rails Inc consisting of a large group of core committers would have an unfair advantage in the training and consulting space - easily siphoning off all the best juice and leaving little for anything else. There are plenty of examples in our industry of that happening around open source tools.

It's much more satisfying to see a broader pool of companies all competing on a level playing field.

Disclaimer: In the past, I've provided training and consulting around Spring - in addition to writing a book about it. Interface21 has never done anything to discourage people from using my services. At least they haven't done anything that I know of.

GlassFish 2 vs. Tomcat 6

In Switched, Dave says:

Now that Glassfish V2 is out I'm switching from Tomcat to Glassfish for all of my development. It's more than fast enough. With Glassfish on my MacBook Pro, Roller restart time is about 8 seconds compared to 16 with Tomcat. And the quality is high; the admin console, the asadmin command-line utility and the docs are all excellent. The dog food is surprisingly tasty ;-)

I did some brief and very non-scientific performance comparisons myself:

Startup Time with no applications deployed:

• Tomcat 6: 3 seconds
• GlassFish 2: 8 seconds

Startup Time with AppFuse 2.0 (Struts + Hibernate version) as a WAR

• Tomcat 6: 15 seconds
• GlassFish 2: 16 seconds

Environment:

• JAVA_OPTS="-Xms768M -Xmx768M -XX:PermSize=512m -XX:MaxPermSize=512m -Djava.awt.headless=true -XX:+CMSClassUnloadingEnabled -XX:+CMSPermGenSweepingEnabled -XX:+UseConcMarkSweepGC -server"
• OS X 10.4.10, 2.2 GHz Intel Core 2 Duo, 4 GB 667 MHz DDR2 SDRAM

Since this was a very non-scientific experiment, it's possible the last two are actually the same. It's strange that Dave is seeing Roller startup twice as fast on GlassFish. Maybe they've done some Roller deployment optimization?

I realize startup times aren't that important. However, as Dave mentions, they (and context reloading) can be extremely important when developing.

Update: I got to thinking that Dave is probably referring to context reloading. Here's a comparison of how long it takes for both servers to pick up a new WAR (and start the application) when it's dropped into their autodeploy directories.

• Tomcat 6: 14-16 seconds
• GlassFish 2: 9 seconds

The strange thing about Tomcat is it takes 6-8 seconds to recognize a new WAR has been deployed. Does Tomcat have a polling increment that can be increased during development?

Regardless, it's impressive that the GlassFish guys have made things that much faster for developers. Nice work folks!

These days, I try to use mvn jetty:run on projects. Then I don't have to worry about deploying, just save and wait for the reload. Time to wait for AppFuse 2.0 to reload using the Maven Jetty Plugin (version 6.1.5)? 7 seconds. Of course, it'd be nice if I could somehow get this down to 1 or 2 seconds.

Maybe Dave should use the Maven integration for Roller to decrease his reload times. ;-)

Proposed Tomcat Enhancement: Add flag to escape JSP's EL by default

I posted the following to the Tomcat Developers mailing list. Unfortunately, it didn't get any responses, which means (to me) that no one cares about this feature. I guess the good thing is they didn't veto it.

Hello all,

I'm working for a client that's using a proprietary Servlet/JSP-based framework that runs on Tomcat. They have their own custom JSP compiler and they're looking to move to a standard JSP compiler. One of the things their compiler supports is automatic escaping of XML in expressions. For example, ${foo} would be escaped so <body> -> &lt;body&gt;. JSP EL does not do this. It *doesn't* escape by default and instead requires you to wrap your expressions with <c:out/> if you want escaping.

I'd like to ask what developers think about adding a flag (similar to trimSpaces in conf/web.xml) that allows users to change the escaping behavior from false to true?

I think this is a good option to have as it allows security-conscious organizations to paranoid and escape all content by default.

Thanks,

Matt

Related: http://raibledesigns.com/rd/entry/java_web_frameworks_and_xss

What do you think? Is there anything wrong with adding this (optional) feature to Tomcat? Enhancing security is a good thing - right?

Update: I've entered an enhancement request for this feature and attached a patch.

AppFuse 2.0 Released!

I'm extremely happy to announce we've finally finished developing AppFuse 2.0. The road to AppFuse 2.0 has been a long journey through Mavenland, annotations and generics. Thanks to all the developers, contributors and users for helping test, polish and prove that AppFuse 2 is an excellent solution for developing Java-based applications. Your time, patience and usage of AppFuse has made it the strong foundation it is today. Last but certainly not least, thanks to all the great Java developers who wrote the frameworks that AppFuse uses - we're truly standing on the shoulders of giants.

What is AppFuse? Click here to find out.

AppFuse is an open source project and application that uses open source tools built on the Java platform to help you develop Web applications quickly and efficiently. It was originally developed to eliminate the ramp-up time found when building new web applications for customers. At its core, AppFuse is a project skeleton, similar to the one that's created by your IDE when you click through a wizard to create a new web project.

AppFuse 1.x uses Ant to create your project, as well as build/test/deploy it. AppFuse 2.0 uses Maven 2 to create your project as well as build/test/deploy it. IDE support is much better in 2.0 because you can generate the IDE project files with Maven plugins. AppFuse 1.x uses XDoclet and JDK 1.4+.

This project was started in April 2003.

AppFuse 2.0 is available as a Maven archetype. For information on creating a new project using this release, please see the QuickStart Guide or the demos and videos.

If you've used AppFuse 1.x, but not 2.x, you'll might want to read our Frequently Asked Questions. If you have any questions or issues, please post them to the user mailing list. The Maven Reference Guide has a map of Ant » Maven commands. Maven for Newbies might also be useful if you've never used Maven before. There is some support for Ant in this release.

AppFuse 2.0 contains over 200 pages of documentation, downloadable as a PDF (3 MB). You can also download all its dependencies and install them in your local repository if you want to work offline.

For more information, please see the 2.0 Release Notes. The 2.0 series of AppFuse has a minimum requirement of the following specification versions:

• Java Servlet 2.4 and JSP 2.0 (2.1 for JSF)
• Java 5+

New features in AppFuse 2.0 include:

• Maven 2 Integration
• Upgraded WebWork to Struts 2
• JDK 5, Annotations, JSP 2.0, Servlet 2.4
• JPA Support
• Generic CRUD backend
• Full Eclipse, IDEA and NetBeans support
• Fast startup and no deploy with Maven Jetty Plugin
• Testable on multiple appservers and databases with Cargo and profiles

We appreciate the time and effort everyone has put toward contributing code and documentation, posting to the mailing lists, and logging issues.

We're also grateful for the help from our sponsors, particularly Atlassian, Contegix, JetBrains, and Java.net. Atlassian and Contegix are especially awesome: Atlassian has donated licenses to all its products and Contegix has donated an entire server to the AppFuse project. Thanks guys - you rock!

Comments and issues should be posted to the mailing list.

Does becoming a committer hurt?

I've recently been offered the opportunity to become a committer on one of the open source web frameworks I often talk about. From a professional and career standpoint, I'd be a fool to turn it down. My clients would benefit and it would look good on my resume. However, I'm wondering if it would hurt how folks perceive me when I talk about web frameworks?

Personally, I don't think it would change my views. Instead, it would probably eliminate my complaining about said framework because its developers would say "If you have a problem with that - fix it!" What do you think - if I were to become a committer on one of Java Web Framework projects, would you still think I'm agnostic when I talk about Java (or other open source) web frameworks? Of course, it's possible you don't think I'm agnostic now, so maybe I'm worried for no reason.

Made it to Helsinki!

After a long couple days of flying, we finally arrived at our hotel in Helsinki this evening. It was a pretty tiring experience getting over here, but that's my own fault since I pulled an all-nighter on Friday night (trying to finish up a training course and slides for CSS). I rewarded myself with a little first-class action from Denver to Newark, but then was in the "back of the bus" from Newark to Oslo. We had a 4-hour layover in Oslo before hopping on a 2-hour flight to Helsinki. A 45-minute cab ride, a stop at the wrong hotel, a 5 minute cab ride and we arrived at the right location. I should sleep pretty good tonight after 48 hours of little to no sleep.

We haven't seen much of the city yet, but hope to tomorrow. The hotel we're staying at is pretty nice. I especially like the fact they have very nice saunas and we're staying only a few doors down from them. Our room is incredibly large and has 3 beds in it. The bathroom even has heated floors. Good thing my mom picked up the tab for this one.

Tomorrow looks like a high of 55°F, which sounds about perfect. I brought a coat and would hate to pack it around for nothing.