ColdFusion to Railo. Interesting Issue with Flex Remoting

I was trying to figure out a solution to a minor problem that we recently ran into with our project. In brief, we have a CFC that communicates, as a RemoteObject, with our Flex front-end. One of the methods in this CFC receives an array of objects and must store the object data in our database. There are a few scenarios where this array of objects could contain duplicates. After some discussion, we opted to have the back-end (CFML) check for and remove duplicate objects, should they exist, from the array.

This is a task I've often had to do with lists, which is quite simple with CFML alone. But with these array of objects, I had to verify that each property in the object matched another object. I considered a solution wherein I looped over the array of objects and performed the necessary comparisons on each property, and so on down the line.

While I knew I could make it work, everything I thought of felt like too much code and unnecessarily cumbersome/complex. I kept thinking that there had to be a way to use Java to help myself out. I am always looking for ways I can 'sneak' a little Java into my applications. However, not being an accomplished Java programmer (to say the least), it's really tough to pull off!

Anyway, I did a little research on removing duplicate objects from arrays in ColdFusion and found a really helpful post from Adam Presley (http://blog.adampresley.com/2009/removing-duplicate-array-items-in-coldfusion/) that got me started. Adam's post and code snippets worked great when I sent in an array of simple objects (strings, integers, and the like) but I wasn't able to remove duplicates when I passed in the array of objects. Even though it did not directly solve my problem, Adam's post helped me in that I was able to see a clear and related example of how I could use Java to my advantage in a CF project.

In terms of using Java in ColdFusion, I (too) often forget that you can just call native Java methods on ColdFusion objects. However, I am not well-versed in what those methods might be for a specific ColdFusion object. I did a little Googling and found a handy post from Christian Cantrell on how you can tell what data types a ColdFusion object inherits from Java. As noted in his post, a ColdFusion array inherits from the java.util.Vector. A 'quick' perusal of the Java API docs and I learned of two important, relevant methods available to Vectors: add and contains. 

Contains and add do exactly what you might expect. Contains checks the array to see if it already contains the passed in object and returns a boolean. Add will add the object to an array. These methods are incredibly helpful and here's how it solved my problem. 

function removeDuplicatesFromArrayOfObjects(aryOfObjects){   
  var deduped_array = [];      
  for(i = 1; i < arrayLen(arguments.aryOfObjects); i++){
    if( !deduped_array.contains(arguments.aryOfObjects[i]) ){
      deduped_array.add(arguments.aryOfObjects[i]); // it's so cool how you just call .add on a CF array!
    }
  }   
  return deduped_array;
}

One thing I noticed is that the contains method evaluated objects as equal regardless of the casing of a string. For example, I had two objects where one string property was "A" and another was "a". The contains method saw them as the same. There may (probably?) be a way to prevent this behavior and treat strings with differing cases as unique objects but I honestly don't know enough ... yet ;).

I haven't quite mastered Posterous's syntax highlighting and don't think it supports ColdFusion, so I converted my CFML tag-based function to a script-based one (and in the process may have messed something up ... apologies if that's the case but it seems ok).

I know this isn't an earth-shattering revelation or the most kick-ass use of ColdFusion's Java prowess but it saved me a good bit of time and for those, like me, who are often unsure of how you can leverage Java in CFML, I wanted to post my experience. Plus, it's only six freaking lines of code to find duplicate objects in an array of objects!!

Posted via web from Craig Kaminsky's posterous

While running various performance tests (load balancing, DB drivers, etc.) on an app we're porting to Railo from ACF 8.0.1, I decided to try out the newest JDBC driver from Microsoft for SQL Server 2008 (version 3.0 of this driver was released in April 2010), as the built-in MS SQL server driver notes that it is for SQL Server 2005 (or so say my error logs). 

I had yet to use the "Other" option in the Railo Web Administrator for a datasource and wanted to record what steps I took for later (or in the event someone else might be interested in the same). The process itself is pretty simple:

1. Download the newest driver from Microsoft: http://www.microsoft.com/downloads/details.aspx?FamilyID=%20a737000d-68d0-453...
2. Unpack/unzip the downloaded file. For me, expanding the archive resulted in a folder called sqljdbc_3.0 
3. Inside the expanded folder is another, called enu (or whatever language was selected with the download from step 1), which contains, among other items, a file called sqljdbc4.jar
4. Copy or move sqljdbc4.jar to your Railo 'install' folder (I use Railo on Tomcat and, for me, this is at /Developer/tomcat/railo) ... it'll be the folder that contains a railo.jar file :)
5. Restart Railo
6. Open the Railo web administrator for your app
7. Go to the Datasource page and enter your datasource name, selecting "Other - JDBC Driver" from the Type drop-down
8. Enter your username and password as required by your DB on the ensuing page ('Create new datasource connection Other - JDBC Driver')
9. For the Class field, enter: com.microsoft.jdbc.sqlserver.SQLServerDriver
10. For the DNS field, enter: jdbc:microsoft:sqlserver://SERVERNAMEORIP:1433;DatabaseName=DBNAME

That's it. Not much to it, I know. However, I was not sure of what the right class name was for the MS driver and how to create the connection string for the DNS ... and, as I mentioned above, wanted to note it for future reference.

For anyone wondering, the performance of Railo's built-in Microsoft driver (vendor Microsoft) was nearly identical to the newer JDBC 3.0 driver. I used a pretty intensive query (lots of calculations, etc.) as my unofficial test case. The connection via the 3.0 driver was typically, but not always, a few milliseconds faster. When it was faster, it truly was negligible ... we're talking differences such as 437 ms for the 3.0 driver to 454 ms for the built-in driver.

Even though there wasn't any true performance gain, it's good to know how to deploy and use other JDBC drivers!

Posted via email from Craig Kaminsky's posterous