April, 2010

...now browsing by month

 

Why I use Linux as a desktop

Thursday, April 22nd, 2010

Penguins are cool!

I have two Macs, a high-end PC running Linux, my work laptop running Linux, three PCs running Windows XP, and one more laptop also running XP in my house. I have several PCs that are mostly used for parts. An upper end PC running Linux that is my main server that is my file server on the host level, and runs a web server, mail server, database server, test server all as virtual machines. Lastly, I have a left over PC that I am using as a special project database server for APEX applications that is also running Linux.

I have told the kids that as each of their PCs die, they will be replaced by a Mac Mini. The one that I currently have is about five years old, and the only thing I do is turn it off before leaving the house on vacations. I don’t think it has ever crashed. Ever!

So why not use a Mac as my primary platform. Cost for one. I can get much more horsepower for a lot less money with Linux. This year for my Christmas present to myself I built my own dream machine. I already had an LCD monitor, keyboard and mouse and was limited to about $500 dollars. It is amazing what you can purchase at Newegg! I was able to get a 3.2 GHz Quad core processor, 8 GB of high speed ram, and a fast SATA HD. I also replaced the stock fan, just in-case I ever decide to overclock the CPU to 4 GHz or higher. Sometimes just knowing you can is just as good as actually doing.

I still haven’t answered the question. I spend most of my time using ant to build java projects, JDeveloper as an IDE, and SQL Developer working with PL/SQL. Linux is simply so much faster, especially starting up SQL Developer or JDeveloper. Even on my work laptop (which is painfully slow to begin with), I can launch SQL Developer and have the IDE ready to go in 15 seconds, where my colleagues can wait more than a minute, sometimes two – just to see the IDE. I don’t even want to talk about my PPC Mac. Apple doesn’t have a version of Java that will run any of the current version of JDeveloper or SQL Server. Too bad, so sad!

If speed was the only issue, I probably would be running just about everything from the command line as well, but I do have my limits. I love the interface on the Mac. Plain and simple, it is just beautiful. A work of art! I just wish the delete key worked like the backspace key. Delete just doesn’t work right and is really my only complaint. If the Mac was faster and cheaper, I would replace all of my PC’s with them.

Windows. Well, besides re-imaging each PC yearly due to a variety of reasons, blue screens of death when playing games in my free time (ha ha), dealing with viruses that caused another round of re-imaging, etc… I like it. My dream includes buying Windows as a Windows manager that runs in X-Windows like Gnome or KDE. Yea, I would buy start buying Windows again if that happens.

I really don’t have any love for the Window managers in Linux. Just like I really don’t have any particular preference for a specific distribution. Typically, I run Fedora on my Desktops, and Ubuntu LTS for all of my servers and virtual machine servers. I have run Gentoo, and think that is a really great distribution. It just is a lot of work to setup, or at least I think it is. Don’t get upset, one of my favorite coffee mugs is my Gentoo Linux mug, and I don’t have any other Linux mugs.

And what’s wrong with Linux. I can’t run notepad++ or textpad natively. I really dislike vi and emacs even more so. Windows has the best text editors of all the platforms. That will hopefully be fixed one day! Until then, I’ll happily type away using gedit. And yes, I still use Windows running in VMWare for Visio and Microsoft Project. As long as these products run under XP, I will keep waiting for Windows as a Window Manager.

How to validate against a system table when checking in a document

Thursday, April 22nd, 2010

One of the problems we continue to face is comboboxes that contain a lot of data. 40,000 rows of data. Just having a few of these on a form can really make the system slow enough to be unusable. One solution is to just validate the data on submit. The key is to not loose the data if there are any problems.

Using examples from the howtobundle of examples from Oracle and also using Bex’s book – Custom Java Components chapter, I have put together a quick example. First create a component in the ComponentWizard. I named mine EmployeeCheckinValidation. Once you have this created (Follow the directions in Bex’s book if you are having trouble) create a project in JDeveloper. For this to work, you have to use the same JDK that is used in your UCM instance. In my case, I created a Samba mount point, and was able to point to the correct JDK using this method. You also have to add UCM’s server.jar file to your project’s jar library. To make things easier, I also set the projects source directory and output directories to directories named src and classes in the component – in my example ucm/server/cusom/EmployeeCheckinValidation/src and ucm/server/cusom/EmployeeCheckinValidation/classes. The last change was to the hda file. I added the following line below the line containing the version:
classpath=$COMPONENT_DIR/classes

Now when I rebuild the project, both the src and classes get updated automatically. Don’t forget to restart your server anytime you rebuild your file!

Here is the sample file that I have created. Basically, this will check that the title is not in the list of doctypes, which is one of the system tables. I’m sure there are other ways to accomplish this, but I thought I would share one way to do this. If you have ideas to make this better, please leave me a comment!

package testing;

import intradoc.common.ExecutionContext;
import intradoc.common.ServiceException;
import intradoc.common.SystemUtils;

import intradoc.data.DataBinder;
import intradoc.data.DataException;
import intradoc.data.DataResultSet;
import intradoc.data.ResultSet;
import intradoc.data.Workspace;

import intradoc.provider.Provider;
import intradoc.provider.Providers;

import intradoc.shared.FilterImplementor;

import intradoc.util.IdcMessage;


public class CheckinFilter implements FilterImplementor {
    public int doFilter(Workspace ws, DataBinder binder, ExecutionContext cxt) throws DataException, ServiceException {
        SystemUtils.trace("system", "Starting doFilter for EmployeeCheckinValidation");
        
        // Display the binder
        // System.out.println(binder);
        
        String dDocTitle = binder.getLocal("dDocTitle");
        SystemUtils.trace("system", "dDocTitle=" + dDocTitle);

        String SQL = "select count(*) counter from doctypes where upper(ddoctype) = '" + dDocTitle.trim().toUpperCase() + "'";
        String ResultSetName = "DDOCTYPECOUNT";
        SystemUtils.trace("system", "SQL=" + SQL);
        
        // Search database for ddoctitle in doctypes.ddoctype  Fail if
        // this this is true
                
        if (ws == null) {
            SystemUtils.trace("system", "ws is null, getting ws from call to getSystemWorkspace()");
            ws = getSystemWorkspace();
        }
        
        String value = "";
        int ivalue=0;
        DataResultSet result = null;
        DataException error = null;
        SystemUtils.trace("system", "try-catch block to get ResultSet from SQL");
        try {
            ResultSet temp = ws.createResultSetSQL(SQL);
            result = new DataResultSet();
            result.copy(temp);
            binder.addResultSet(ResultSetName,result);  // Makes results avilable for other Java methods or IdocScript templates.
        } catch (DataException de) { 
            error = de;
        } finally {
            // ws.releaseConnection();  // do NOT uncomment this, unless you like broken code!
        }

        try {
            result.first();
            // value = result.getStringValueByName("counter");
            value = result.getStringValue(0);
            SystemUtils.trace("system", "value(getStringValue(0)) = " + value);
            ivalue = Integer.parseInt(value);
            SystemUtils.trace("system", "ivalue=" + ivalue);
        } catch (NumberFormatException nfe) {
            SystemUtils.trace("system", "nfe.getMessage()=" + nfe.getMessage());
        }

        
        if (ivalue > 0) {
            SystemUtils.trace("system", "Throw error since dDocTitle cannot be a dDocType");
            throw new ServiceException("dDocTitle cannot be a dDocType");
        }
        
        if (error != null) {
            SystemUtils.trace("system", "Throw error since error condition exists.");
            throw error;
        }

        SystemUtils.trace("system", "Ending doFilter for EmployeeCheckinValidation");
        return CONTINUE;
    }
    
    public Workspace getSystemWorkspace() {
        Workspace workspace = null;
        Provider wsProvider = Providers.getProvider("SystemDatabase");
        if (wsProvider != null) {
            workspace = (Workspace)wsProvider.getProvider();
        }
        return workspace;        
    }
}

Comments

Friday, April 16th, 2010

I have enabled comments. Now don’t go crazy, as they are moderated – I have to read and approve them before they show up to everyone.

I hope to stick to a basic rule: Spam and inappropriate language goes to file 13, everything else stays.

Apache AXIS 2

Friday, April 16th, 2010

Two years ago (2009), a colleague of mine and I presented at RMOUG on using Oracle’s BPEL with Microsoft’s SQL Server to create Web services. Since we were using SQL Server instead of Oracle, we had to create xsd files to define our data. This took quite a bit of time, though not nearly as much time as dragging arrows to connect each of the items in the xsd file. In the end, we had a working deployment that realistically took about six months of duration and maybe 300 hours of work. Most of the problems were due to lack of available documentation at the time, and a difficult time getting Oracle’s BPEL product to play nice with SQL Server. The same service using an Oracle database could be done quite fast. Once you get the hang of it, we are talking minutes. Of course testing takes quite a bit more time.

Well, fast forward to the present. I was in a meeting and the topic of using Web services in Application Express was bantered about. After a few minutes, most of the team was looking at me. I am the one who experiments with products, and they were correct in assuming that I had the solution. Actually, there is a weather example available at Oracle. I was able to demo this example in a few minutes, and show how easy Web services are to integrate into APEX. The conversation then moved on to discussing whether the organization should use our purchased SOA suite to build a Web service library, but that is an entirely different story.

I found myself unable to stop thinking about the difficulty experienced using SQL Server and BPEL. I really couldn’t stop thinking about it. So, I did something about it. Several years ago, I had written a paper on Web services using AXIS. This had been several years ago and AXIS was now using a completely rewritten code base. What’s a guy to do? After downloading the Axis2 1.5.1 Release, I extracted and deployed the included war file to the Tomcat server running on my laptop. There is a samples directory, and one of the easiest ways to get started is the pojoguide example. Basically, there are four files that make up this sample. There is the service definition, src/META-INF/services.xml. In this file, you name the web service that is defined in the service directory.
In the data directory is a bean, src/pojo/data/Weather.java, which simply defines the output. For simple Web services in AXIS2, it seams easy to just create a class, and either return this class, or an array of the defined class as output. Believe me, this method is really easy to use! Okay, the rpcclient is more complicated. In my case, I replace this with a client directory, and created a Test.java file. I use this to test my Web service. After all, it is really just a Java class. In the end, I simply use a browser to test my Web service, displaying the results in my browser. The big deal is the data in the service directory. src/pojo/service/WeatherService.java contains the actual Web service. The class is the Web service, and all methods can be services that you call. Easy cheesy! For examples, I have just copied the pojoguide directory and simply renamed it, then updated the four classes described above. Of course, you have to update the build.xml file to point to the correct files and directories, but that isn’t too bad. Use ant to build the Web service to generate deployable file and test the service itself.

So back to my original point. After downloading the jtds SQL Server driver from SourceForge, I copied the jar file to the lib directory under axis2/WEB-INF and restarted Tomcat. I also had to add this to my build path in ant to get an error free compile. Now I could connect to my database. 15 Minutes later I had created my first AXIS2 Web service in years. Another ten minutes, and I was able to integrate this into APEX. Two great products working together. Technology is awesome!