For prototyping purposes I usually prepare small datasets. With CCC you can just prototype everything in a standard HTML file (which requires no connection to the Pentaho BA Server or any database): Here we just define an inline JSON style dataset (well, actually a JavaScript object to be precise). This works extremely well. This inline dataset looks exactly the same as the resultset a CDA query would return.

In CDE we have the Scriptable data source, which allows you to create an inline dataset using Beanshell (Java for the ordinary folks). This feature is mainly provided by the PRD libraries, so it works in CDE as well as in Pentaho Report Designer. The big drawback, however, is that the declaration is very verbose and will alienate anyone not familiar with Java.

Another point that interfered with my workflow was that I could not just copy the inline dataset from my CCC testing files to CDE (to continue prototyping there). This was the reason why I created the Scriptable JavaScript Data Source Jira improvement case. And this request was implemented very quickly, this is why I write this article about this brand new feature.

We will build CDA from source as this feature is currently (2015-03-18) only available in the master branch. Execute the commands below (this assumes you have all the build tools installed):

git clone https://github.com/webdetails/cda.git
cd cda 
ant clean-all resolve publish-local

This will take some time to run. Brew yourself a fresh cup of coffee in the meantime.

Once finished, you should see this close to the end of the log output:

BUILD SUCCESSFUL

Next run:

cd cda-pentaho5
ant clean-all resolve dist

Now we can add it to our developement server (this reads: do not use this in production for now!):

1. Delete cda folder in <ba-server>/pentaho-solutions/system/
2. Copy <cda-source>/cda-pentaho5/dist/cda-TRUNK-SNAPSHOT.zip into pentaho-solutions/system/ and unzip it.
3. Restart the BA-Server and you are done!

In CDE under Data Sources Panel > SCRIPTING Queries > jsonScriptable over scripting we define the inline data source like this:

Set the Language in the Properties panel to beanshell and the Query to:

{
	"resultset": [
		["2006 Q1", 242],
		["2006 Q2", 410],
		["2006 Q3", 340],
		["2006 Q4", 353]
	],
	"metadata": [
		{ "colIndex": 0, "colType": "String", "colName": "Year-Quarter" }
		,
		{ "colIndex": 1, "colType": "Numeric", "colName": "Value" }
	]
}

And that’s it. You can now use this data source in your dashboard! Extremely useful for prototyping!