Monday, February 28, 2011

Tweaking Google Code's Issues Gadget

I'm a big fan of Google Code where I host the littleware project. I've been playing around with gcode's issue tracker trying to figure out some way to embed the tasks associated with a code sprint into the sprint's overview documentation, and I eventually stumbled upon a nice issue-tracker gadget in a Gadgets for Google Code bLog entry that almost does what I want, but not quite!

Fortunately, it turns out that the issue-tracker gadget is one of the gadgets developed by google's code-hosting team, and the code is open source here. I cloned a copy of their issue-tracker gadget under a littleware 'gadgets' repository, and was able to add a milestone parameter to the gadget with only a little bit of trouble. I think I'm all set!


I submitted an issue with the gadgets for google code project with some information about my patch. It would be cool if they fold the patch into their code - we'll see what happens.

Wednesday, February 23, 2011

Success tastes like pizza!

I think I finally have the blogger podcast thing working. To recap - I started by hosting my .mp3 media file with Google docs, and that works great, except Google docs publishes the .mp3 file to a URL that does not include the .mp3 file extension (something like https://docs.google.com/uc?id=0B-82pgvnWy-8OWM2ZWJhYzktOWU5OC00MTI4LWFiZWYtZWNhOTQ0ZjlmMDA2&export=download&hl=en), and iTunes requires that a podcast's enclosure references a URL with a valid media extension (more details here).

So I went looking for some other free place to host my .mp3 files, and found Microsoft's sky drive, which preserves a file's extension in its download URL (ex: http://public.blu.livefilestore.com/y1p5DNvAssqO8iaDVAKdFChzS01Pj_v3M9B68-VJdTj9sQr43GHTr8pThRglPvVeIzJlbU4AE3nQbW7NiocbkkOLA/shCore.js?download&psid=1). Unfortunately, that download URL does not persist - a file's URL changes from day to day.

Finally, I tried Google sites, and (so far) I think that works - with persistent iTunes-friendly URLs like https://sites.google.com/site/frickjack/projects/podcasts/Podcast20110223.mp3?attredirects=0&d=1

This picture shows our dog, Moose, guarding a pizza from our other dog, Ponzi - Moose wants it all for himself! One tip for home made pizza - if you don't have a pizza stone, then an upside down iron skillet works too.

From Food
Picasa logo

Skydrive lets me down!

Ugh!

I was pretty happy when I was able to setup a blogger based podcast with mp3 files served from Microsoft's sky drive, so I went ahead and linked my blogger template to a copy of Gorbachev's syntax highlighter on sky drive too, but it turns out the freakin' sky-drive URL's do not persist, so neither of those nifty tricks work. I just inlined the syntax highlighter into the blogger template, but I'll have to figure something else out for audio-file hosting. There must be some way to get Google Docs URL's to work with iTunes.

This is from Apple's "Making a Podcast" page:
The URL before the GET-style form values (before the first ?) must end in a media file extension (e.g. mp3). To work around this, the feed provider can alter their URL from this: http://www.podcaster.com/load.php?f=&Wipeout.php to this: http://www.podcaster.com/load.mp3?f=&Wipeout.mp3 Notice how it says load.mp3 instead of load.php. It should be possible to accomplish this via various means, such as web server rewrites. iTunes looks at the extension of the path part of the url, i.e. the part before the"?".

Ugh!

JScrollPane a JPanel !

If you're a frickjack like me, then you might believe that a JScrollPane only works with widgets like JTextArea that are specially designed to be scrollable, but (as with so many other beliefs) you would be wrong! Turns out JScrollPane works just fine with JPanel, which is a big deal when you're trying to layout a bunch of widgets on a big form or whatever. Anyway, the scala code below works, so I hope scrolling works for more complicated panels too. I'll let you know if I find out otherwise.

package littleware.demo

import javax.swing._

object Toy {
    def main( args:Array[String] ):Unit = {
        SwingUtilities.invokeLater( new Toy )
    }
}

class Toy extends Runnable {
    override def run():Unit = {
        val jbox = new Box( BoxLayout.Y_AXIS )
        (0 until 40).foreach( (index) => {
            jbox.add( new JLabel( "Test" + index ) )
        }
        )
        val jframe = new JFrame( "Toy" )
        jframe.setDefaultCloseOperation( JFrame.EXIT_ON_CLOSE )
        val scroll = new JScrollPane( jbox )
        scroll.setVerticalScrollBarPolicy( ScrollPaneConstants.VERTICAL_SCROLLBAR_ALWAYS )
        scroll.setPreferredSize( new java.awt.Dimension( 700, 400 ))
        jframe.add( scroll )
        jframe.pack
        jframe.setVisible( true )
    }
}

How do you like that slick syntax highlighting ? If it works (I won't know till I post, so frick), then it's all thanks to Alex Gorbachev's syntax highlighter - which I copied over to Skydrive, and hooked into the bLog template. Alex is probably an evil James Bond villain, but I'll have to paypal him $10- when I get motivated.

That's all pretty exciting stuff. Our dogs were very excited too ...

the pack is bored

Friday, February 18, 2011

Pound cake!

 

What to do with 1/2 pound butter, 5/3 cups sugar, 2 cups flour, ... Good old Fannie Farmer!

 

I also accidentally uploaded this pretty cool picture of my dog, Shug, while playing with Picasa. Go, Shug!

Posted by Picasa

Thursday, February 17, 2011

Blogger + Google Docs Podcast

I often enjoy listening to podcasts while I'm playing around with the computer. The java posse, TED, Entrepreneurial Thought Leaders, and the New Yorker Political Scene are some of my favorites.

I recently began wondering how I could publish my own podcast of (for example) a lecture at the university or whatever with as little expense and effort as possible. It turns out that cheap and easy podcast publishing can be a simple 3 step process thanks to great free online services like Google's docs and blogger, and the very useful open source audacity audio program (or Garage Band is probably even better on the Mac).

  • First, record your podcast with audacity or garage band or your iPhone or whatever. You can get as professional as you want - I'm just a hack for now.
  • Second, publish your audio file online. Google docs gives you 1 GB of storage for free, but Microsoft's skydrive provides 25 GB for free!
  • Third, setup a podcast feed that links to your audio file. A podcast feed is just a specialized RSS feed, which is just an XML file that readers know how to interpret. You could just edit an rss xml file, and publish that via Google docs or skydrive, but Google's blogger platform makes it easy to associate audio files with blog entries, so the blog feed also acts as a podcast feed.
  • Finally, register the podcast's feed url (ex: http://blog.frickjack.com/feeds/posts/default?alt=rss) with iTunes or your favorite podcast player.

Even an idiot like me can pollute the internet with worthless media!

Update ...

I wrote this post before I actually had the podcast feed working, and I wound up driving myself crazy trying to get iTunes to recognize the podcast when hosting the .mp3 file on Google docs. The feed worked fine with Google Reader, but I think iTunes looks at the enclosure URL to decide what kind of media file a feed item references rather than trusting the enclosure's mime type attribute. Google docs generates nonsensical urls that do not preserve the file extension (ex: https://docs.google.com/uc?id=0B-82pgvnWy-8OWM2ZWJhYzktOWU5OC00MTI4LWFiZWYtZWNhOTQ0ZjlmMDA2&export=download&hl=en is a .mp3 file), so iTunes throws a
There are no playable episodes ... The URL might point to text-only episodes, or contain file types that iTunes cannot play.
error when loading the podcast.

Fortunately, Skydrive came to my rescue with an iTunes friendly URL for my audio file: http://m8cb3q.blu.livefilestore.com/y1pCrRuW5NqZivMGgB7_oAUy_g8m_83domL0-7pAkoVezoGWOATDPuuLQNxECsNAK60JSxlYwkgHW2WycExfTMGXBxd4-gM0bqR/Podcast20110217.mp3?download&psid=1.

It also turns out that Google Reader is nice about letting random frickjacks embed its audio player!

Another Update ...

Freakin' sky-drive url's don't persist (there's some kind of cache timeout mojo), so the .mp3 links are now bogus. Back to the drawing board!

Wednesday, February 09, 2011

Apache mod_proxy build trickery

I lost a few more hairs building mod_proxy for an existing Apache install. It turns out mod_proxy requires proxy_util.c on the apxs build line.

   apxs -i -c mod_proxy.c  proxy_util.c

Of course the .so builds and installs fine without proxy_util.c, so you think everything is fine until apache pukes with a "missing symbol" error. Frickjacks!

Friday, December 03, 2010

Chrome RSS extension

Before today when I wanted to add a web site's feed to Google Reader from Chrome I would "view source", find the feed URL, then go to google reader to subscribe to the feed. Today I finally got a brain, and checked to see if there's a chrome extension that automates this process, and I found exactly what I wanted. . It's great when someone solves a problem for me!

Tuesday, November 23, 2010

Glassfish JDBCRealm authentication against DSpace user database

A project we're working on at the library requires a DSpace repository webapp to cooperate with a custom data-submission webapp that we're building. We want our custom webapp to authenticate with the same credentials that dspace manages in its internal database. Luckily, our glassfish app-server includes a JDBC security realm that fit this problem perfectly.

It took me several hours of head banging and googling ( http://blogs.sun.com/swchan/entry/jdbcrealm_in_glassfish_with_mysql , http://blogs.sun.com/swchan/entry/jdbcrealm_in_glassfish , http://blogs.sun.com/swchan/entry/assign_groups , http://stackoverflow.com/questions/719708/java-web-application-using-a-custom-realm ) to figure out all the steps to properly configure a glassfish JDBCRealm to authenticate our webapp. I was really glad it worked in the end. I configured a glassfish JDBC security realms with the following properties to access the 'email' and 'password' column's in the dspace 'eperson' database table. The following data in domain.xml can be configured via the glassfish admin webapp

    <jdbc-connection-pool
datasource-classname="org.postgresql.ds.PGSimpleDataSource"
is-isolation-level-guaranteed="false" res-type="javax.sql.DataSource"
description="connection to local dspace database" name="dspace">
      <property name="DatabaseName" value="dspace" />
      <property name="LogLevel" value="0" />
      <property name="LoginTimeout" value="0" />
      <property name="Password" value="XXXXX" />
      <property name="PortNumber" value="0" />
      <property name="PrepareThreshold" value="5" />
      <property name="ProtocolVersion" value="0" />
      <property name="ServerName" value="localhost" />
      <property name="SocketTimeout" value="0" />
      <property name="Ssl" value="false" />
      <property name="TcpKeepAlive" value="false" />
      <property name="UnknownLength" value="2147483647" />
      <property name="User" value="XXXXX" />
    </jdbc-connection-pool>
 
    <jdbc-resource pool-name="dspace" description="dspace data source"
jndi-name="jdbc/dspace" />
 
        <auth-realm
classname="com.sun.enterprise.security.auth.realm.jdbc.JDBCRealm"
name="dspaceRealm">
          <property name="datasource-jndi" value="jdbc/dspace" />
          <property name="jaas-context" value="jdbcRealm" />
          <property name="user-table" value="eperson" />
          <property name="user-name-column" value="email" />
          <property name="password-column" value="password" />
          <property name="group-table" value="epersongroup" />
          <property name="group-name-column" value="name" />
          <property name="assign-groups" value="users" />
        </auth-realm>
 

Finally, we configure our webapp to use the dspaceRealm with the following blocks in WEB-INF/web.xml:

 
    <!--
       Login config for production server.
       See role-mapping in sun-web.xml too.
       Details of dspaceRealm configuration for glassfish in bLog.
 
    <login-config>
        <auth-method>BASIC</auth-method>
        <realm-name>dspaceRealm</realm-name>
    </login-config>
    <security-role>
        <role-name>users</role-name>
    </security-role>
    <security-constraint>
        <web-resource-collection>
            <web-resource-name>Secure Pages</web-resource-name>
            <url-pattern>/*</url-pattern>
            <http-method>GET</http-method>
            <http-method>POST</http-method>
        </web-resource-collection>
        <auth-constraint>
            <role-name>users</role-name>
        </auth-constraint>
    </security-constraint>
   -->

... and WEB-INF/sun-web.xml:

    <!-- Enable on production server - see note in web.xml ....
    <security-role-mapping>
        <role-name>users</role-name>
        <group-name>users</group-name>
    </security-role-mapping>
    -->

In the end I wanted to send an e-mail to the dspace tech-list, but I was disappointed to discover that there is not a standard way to incorporate JAAS authentication into an application server security realm. All the app servers have a slightly different way of implementing security realms, and the setup I describe above is glassfish specific, but that's good enough for me for now.

Saturday, October 02, 2010

Too Big To Fail

Last week I finished reading Andrew Sorkin's Too Big To Fail. It was a great read! I was engrossed by the story of the events behind the financial system's near collapse in September, 2008, and the handful of personalities that worked to save the system. The book especially gives insight into the monumental bad decision to allow Lehman Brothers to go bankrupt, the acquisition or legal transition of the other big investment banks, the AIG mess, the passage of TARP into law, and the government's subsequent capital injection into U.S. banks in exchange for equity.

Sunday, September 26, 2010

Powershell browse and recycle functions

I just added browse and recycle functions to my powershell profile: Documents/WindowsPowershell/profile.ps1. The browse function opens up an explorer window at a specified path, and the recycle function sends a path to the recycle-bin after prompting for user confirmation.

# Excerpt from profile.ps1
...
$GLOBAL:docs = $home + "\Documents"
$GLOBAL:trash = $Env:TEMP
$GLOBAL:shell = new-object -com Shell.Application
$Env:nodosfilewarning=1
...
function recycle( $path ) {
    $clean = resolve-path( $path )
    $item = $shell.Namespace(0).ParseName( $clean )
    $item.InvokeVerb("delete") 
}

function browse( $path ) {
    $clean = resolve-path( $path )
    #echo $clean.path
    $shell.open( $clean.path ) 
}

Sunday, September 12, 2010

Simple Parallelism with Scala

Scala provides some great list processing API's, and it's easy to parallelize Scala's list processing with a java executor service. The following demonstration shows a couple easy ways to parallelize list processing. The second by2Groupwise variation processes chunks of a work list in parallel, and reports progress to the user between chunks. Both by2Groupwise and by2Parallel follow a fork-join pattern where a pass over the work list submits callable tasks to an executor, then a subsequent pass waits for the results.


import java.util.Date
import java.util.concurrent.{Callable,Executors,ExecutorService}
import java.util.logging.{Level,Logger}


object Demo {
    val log = Logger.getLogger( getClass.getName )
    
    implicit def callable[A]( func:() => A ): Callable[A] = new Callable[A] {
         override def call:A = func()
    }
    
    
    val poolSize = 4
    val exec = Executors.newFixedThreadPool( poolSize )
    
    var lastReport = -1;
    def reportProgress( soFar:Int, total:Int ):Unit = {
        val progress = (soFar * 100) / total
        if ( progress != lastReport ) {
            log.log( Level.INFO, "Progress: {0}%", Array[Object]( progress.toString ) )
            lastReport = progress
        }
    }
    
    def timer[A]( work: => A ):(Long,A) = {
        val start = new Date
        val result = work
        ( (new Date).getTime - start.getTime, result )
    }
    
    def by2( arg:Int ):Int = {
         Thread.sleep( 1000 )
         arg * 2     
    }
    
    def by2Sequential( work:Seq[Int] ):Seq[Int] = work.zipWithIndex.map( 
         _ match { case (arg, index) => { 
             reportProgress( index, work.size )
             by2(arg)
           } } )
    
    def by2Parallel( work:Seq[Int] ):Seq[Int] = {
         work.map( (arg) => exec.submit( () => by2(arg) )
                ).map( (future) => future.get )
    }
    
    def by2Groupwise( work:Seq[Int] ):Seq[Int] = {
        val groups = work.grouped( poolSize ).toList
        log.log( Level.FINE, "Frickjack size: {0}", Array[Object]( groups.size.toString ) )
        groups.zipWithIndex.map( 
          _ match { 
            case (group,index) => {
              log.log( Level.INFO, "Scan index: {0}", index )
              reportProgress( index, groups.size )
              group.map( (arg) => exec.submit( () => by2(arg) )
                     ).map( (future) => future.get )
            }
            }
        ).toList.flatten
    }
    
    def main( args:Array[String] ):Unit = {
        val work = 1 until 10
        log.log( Level.INFO, "Seq {0}", Array[Object]( timer( by2Sequential( work ) )._1.toString ) )
        log.log( Level.INFO, "Parallel {0}", Array[Object]( timer( by2Parallel( work ) )._1.toString ) )
        log.log( Level.INFO, "Groupwise {0}", Array[Object]( timer( by2Groupwise( work ) )._1.toString ) )
        exec.shutdown
    }
}

Saturday, August 21, 2010

frickjack.com DNS forwarding to Google Sites

Today I finally wired up my frickjack.com domain configuration with GoDaddy to just redirect (via domain forwarding) to pages hosted on Google Sites. I bought the frickjack.com domain a few years ago with the idea that I'd get a static IP address, and run a web server from home with a bLog, code repostiory, and some demonstration littleware web applications.

I actually did run my own server on an old laptop for a while, but I always felt bad leaving my old laptop running in the corner just to host a site no one looked at anyway. I tried to manage my own bLog with XML and XSL, but Blogger and other hosted solutions are free and easy to use. Similarly, I've been very happy letting Google Code host my littleware project.

Anyway, Google Sites gives me an easy way to put frickjack.com back online as the root node of the graph that connects my disembodied web presence across blogger, facebook, google code, and other sites.

Sunday, June 13, 2010

Bootstrap Bonanza

This post is a copy of this google doc describing littleware's bootstrap process.

I recently reworked the littleware bootstrap and dependency injection process. In the new system Guice manages dependency injection; java's ServiceLoader utility discovers the littleware modules available in the classpath, and OSGi manages startup and shutdown of subsystems. The new bootstrap code also supports different runtime profiles similar in spirit to java EE6 profiles or configurations in Apache Ivy. The new code both simplifies the process an application follows to configure, startup, and shutdown a littleware application; and also simplifies the artifacts that a littleware extension module must provide to register itself with littleware's runtime to expose its functionality to the application.

Littleware has a multi-step bootstrap process. The parent application determines whether to start littleware in a "server", "client", or "app" configuration. With the "client" configuration littleware's bootstrap establishes and authenticates a user session with a remote littleware server (which is another application running a "server" configuration). The "app" configuration allows applications that do not interact with a littleware server to still take advantage of littleware tools that support generic standalone GUI and CLI applications.

The "server", "client", and "app" systems have similar bootstrap processes. A "client" application begins by allocating a ClientBootstrap.ClientBuilder type object:

       final ClientBootstrap.ClientBuilder builder =           
             ClientBootstrap.clientProvider.get();

The ClientBuilder automatically loads the ClientModuleFactory service providers registered in the application class path, and gives the application the opportunity to add or remove modules from the bootstrap module set. The ClientBuilder also allows the application to specify a a profile from the AppBootstrap.AppProfile enumeration. The available profiles include AppProfile.SwingApp (the default), AppProfile.JNDI, AppProfile.CliApp, and AppProfile.WebApp.

The ClientBuilder constructs a LoginSetup object that provides methods to manage user authentication including an automatic method that prompts the user for his credentials in a way appropriate to the AppProfile, and returns a ClientBootstrap object. The ClientBootstrap bootstrap method injects and returns an application supplied type.

        final ClientBootstrap boot = 
               builder.addModuleFactory( new MyModuleFactory() 
                         ).profile( AppBootstrap.AppProfile.CliApp 
                         ).automatic();
        final MyTool tool = boot.bootstrap(MyTool.class);

The littleware runtime is up and running at this point. The MyTool class should be annotated so Guice injects whatever dependences the application needs.

         class MyTool {
                 ...
                   @Inject
                    public MyTool( AssetSearchManager search ) {
                           this.search = search;
                    }
                 ...

Under the bootstrap hood ClientBootstrap loops over the ClientModuleFactory service providers to build a ClientModule (a Guice Module subtype) configured :

         factory.build( profile )

ClientBootstrap builds a Guice Injector configured by the resulting collection of ClientModule objects, then loops over the ClientModule collection again to inject the OSGi BundleActivator associated with each module. Next ClientBootstrap launches an OSGi runtime to startup the activators. Finally, ClientBootstrap injects and returns MyTool.class the application passed to the bootstrap() method.

Saturday, May 15, 2010

Stacked Authentication with JAAS

I was happy the other day to find a quick solution to an authentication problem using java's authentication and authorization service ( JAAS). I was configuring JAAS authentication for a littleware based webapp where I wanted most users to authenticate via active directory, but to also have a simple password file to manage authentication for some internal users like "guest". Fortunately, JAAS easily supports this kind of thing. After googling around I stumbled across the JMX com.sun.jmx.remote.security.FileLoginModule which uses a simple password file, and I was already familiar with using the com.sun.security.auth.module.LdapLoginModule for simple authentication against Active Directory. I simply configured the "littleware.login" context in the Glassfish server domain's login.conf file (below) so that authentication via either A.D. LDAP bind or the password file would both be sufficient to certify a user. Works like a charm!

littleware.login {
        com.sun.security.auth.module.LdapLoginModule SUFFICIENT
             userProvider="ldap://xxxxxx/dc=auburn,dc=edu"
             authIdentity="{USERNAME}@auburn.edu"
             userFilter="(cn:dn:={USERNAME})"
             useSSL=false
             debug=true;
        com.sun.jmx.remote.security.FileLoginModule SUFFICIENT
             passwordFile="/home/glassfish/.littleware/jaasPasswordFile.txt";
};

Tuesday, April 20, 2010

Scala dependent method type

Scala dependent type exception

I'm building a simple JSF-3 web application that implements backing beans in Scala 2.8 One bean includes Enumeration properties that a JSF web-form exposes to clients via a combo-box control configured via a key-value map.

    <h:selectOneMenu id="patronType" value="#{sampleBean.patronType}">
        <f:selectItems value="#{sampleBean.patronTypes}" />
    </h:selectOneMenu>

The sampleBean's definition includes patronType and paronTypes properties which are Enumeration subtypes.

object PatronType extends Enumeration {
  val Student, Faculty, Other = Value
}

...

@ManagedBean
@RequestScoped
class SampleBean extends InjectMeBean {
  private def enumToMap( enum:Enumeration ):java.util.Map[String,Enumeration#Value] =
        ((new ImmutableMap.Builder[String,Enumeration#Value]) /: enum.values) (
      (builder,member) => builder.put( member.toString, member )
    ).build

...
  @reflect.BeanProperty
  var patronType:PatronType.Value = PatronType.Student

  @reflect.BeanProperty
  val patronTypes:java.util.Map[String,Enumeration#Value] = enumToMap( PatronType )
}

I originally defined the patronTypes property as a mapping from String to PatronType.Value.

  @reflect.BeanProperty
  val patronTypes:java.util.Map[String,PatronType.Value] =
               ((new ImmutableMap.Builder[String,PatronType.Value]) /: enum.values) (
        (builder,member) => builder.put( member.toString, member )
      ).build

I introduced the enumToMap method to share that code block between several similar properties with different Enumeration types. Unfortunately - a subtype-specific enumToMap implementation has on output type that depends on an input parameter.

scala> def enumToMap( enum:Enumeration ):java.util.Map[String,enum.Value] =
     |         ((new ImmutableMap.Builder[String,enum.Value]) /: enum.values) (
     |       (builder,member) => builder.put( member.toString, member )
     |     ).build
<console>:6: error: illegal dependent method type
       def enumToMap( enum:Enumeration ):java.util.Map[String,enum.Value] =
                      ^    

I tried several variations on this theme, but finally settled on the enumToMap implementation that just leverages the Enumeration#Value base type. I googled around while working on this code, and stumbled across the fact that scala includes limited experimental support for dependent method types - that will be very cool if it evolves to production status.

Scala's instance-attached nested types have advantages, but they feel alien to an experienced java developer like me.

scala> class A {
     | var a = "A"
     | class B {
     | var b = "B"
     | }
     | var b = new B
     | }
defined class A

scala> val a1 = new A
a1: A = A@6489f0

scala> val a2 = new A
a2: A = A@15c44d6

scala> a1.b = a2.b
<console>:9: error: type mismatch;
 found   : a2.B
 required: a1.B
       a1.b = a2.b
                 ^

Friday, March 26, 2010

Wisdom

I sometimes find myself thinking as I look back at life "Man! I was an idiot back then!" where "back then" might be ten years ago or last year or yesterday or the previous paragraph.

As time passes, I'll think again, "I was an idiot back then" remembering my thoughts the last time I thought what an idiot I was. Eventually I realize that even as I'm thinking what an idiot I was in the past, that in the future I'll look back on the present and think, "What an idiot!".

In time I'll think "I was an idiot when I thought I was an idiot before, because what I did before was not so idiotic although I was too much of an idiot to realize it at the time" - sort of like multiplying two negatives. Of course it's possible to multiply three negatives, "I was an idiot when I thought I was an idiot for thinking I was an idiot".

I know I'll eventually find myself in some bed after the ravages of parkinsons, alzheimers, and multiple strokes leave me with half my body frozen in place and the other half constantly shaking violently. I'll be thinking, "Where am I ? Who is this person feeding me oatmeal ? Who am I ? This oatmeal tastes good! Where am I ? What is this in my mouth ? It tastes good!".

Sunday, February 14, 2010

Larry Ellison in the Cloud

I enjoyed the webcasts Oracle published to outline its product strategy upon the approval of Oracle's acquisition of Sun Microsystems. I was impressed with most of the ideas presented, and I think Oracle will have some success selling their integrated stack "from disk to whatever". However, I was surprised by Larry Ellison's criticism of "cloud computing" when responding to a question near the very end of the event.

Cloud service providers like Amazon EC2 allow a software service developer to deploy one or more copies of a virtual machine image that provides some online software service. A "cloud" service may be a natural evolution of standard hosting provider, but the ability to dynamically allocate and deploy compute resources opens doors for a "software as a service" (Saas) developer. A developer of some service can quickly provision and deploy a virtual server dedicated to some new customer, or scale an application horizontally or vertically in response to increased client load.

A developer team that codes for the cloud avoids the cost and complexity of maintaining servers, networking switches, or storage in a data center. Managing data center complexity is where Oracle makes a lot of its money. Furthermore, the cloud is a natural platform for free open technologies like LAMP that are free to the developer, and do not provide a clear business model for a technology infrastructure companies like Oracle.

Oracle has tremendous breadth and depth of resources, and I don't think Larry Ellison's cloud myopia will be a problem of any importance, but it's strange when a technology leader comes late to an important idea.

Wednesday, January 27, 2010

Apple iPad, Oracle-Sun, and the State of the Union

It's a big day for announcements and media. First, Apple launched the iPad; which looks like it nearly lives up to the hype. Next, Oracle finalized its acquisition of Sun. Finally, President Obama will give his State of the Union address tonight. I wonder which event will be remembered as the most important years from now ?

Sunday, January 24, 2010

Multiple RMI Services on a Single Port

I recently learned of a solution to a problem I've struggled with using java remote method invocation (RMI) to implement client-server applications. I've used RMI for several years, and I think it's a great technology. RMI makes it easy to implement fast and secure remote procedure call between java clients and servers. RMI has two disadvantages compared to REST over HTTP or CORBA as a network computing platform. First, RMI is tied to java, so non-java clients cannot directly access network services via RMI. A second disadvantage of RMI is that RMI exports each of a server's remote objects on a separate TCP network port, so it's difficult to setup firewall rules for an RMI server that dynamically exports distributed objects as different clients request different services.

It turns out that there's an easy workaround to configure RMI to export all of a server's remote objects through a single firewall-friendly network port. I've known for some time that RMI allows server code to specify the port on which to export a remote object. For example, the HelloService below can pass a network port number to its UnicastRemoteObject super class constructor. However, I did not realize that multiple remote objects can specify the same network port to the RMI engine as long as a single RMISocketFactory manages every object on that port (which is the default behavior).

A contrived example below shows a simple scenario where an application dynamically exports multiple remote objects. At boot time, the server initializes a DirectoryService object, and registers the object with an RMI name service. The DirectoryService dynamically creates a HelloService object for each client call to login(). Because DirectoryService and HelloService both specify port 12345 in their constructor, clients access every service object via the same network port.


public class DirectoryService extends UnicastRemoteObject implements Directory {
    public DirectoryService() {
        super( 12345 );
    }

   public login( String name ) throws RemoteObject {
        return new HelloService( name );
   }
}

…

public class HelloService extends UnicastRemoteObject implements Hello {
    private  final String name;

    public HelloService ( String name ) {
        super( 12345 );
        this.name = name;
    }

    public String sayHell() {
        return “Hello, “ + name + “!”;
   }
}