Here are the slides and code from my Adobe MAX ColdFusion Unconference talk 5 Ways To Improve Your App with ORM:

And the code which will set up the Derby DSNs and tables needed.

Available as a download and from Github. After installing:

1. Go to setUp.cfm
2. Enter in your cfadmin password and click 'Make Datasources'
3. Go back to setUp.cfm and click on 'Go to App'

All of the mini apps/examples I used in the demo will now work.

Over the weekend I needed to see what was in my cache. I whipped up a quick script to list everything in cache and from there:

• be able to remove an item
• view the content of an item (limited to top 10 records)
• view the metadata of an item.

I've pretty much overlooked the secondary cache for ORM and, having now played with it, I've realized what a big feature and performance enhancer this is.

In short with secondary cache on ColdFusion can get data from ehCache instead of the database. Set up ColdFusion 9.0.1 and you can share entity data across both ColdFusion instances and servers!

Let backtrack and take a look at a basic set up:

To turn on secondary cache set secondaryCacheEnabled to true (the default is false). In this example I have also turned logsql to true*. Now lets look at our persistent cfc:

User.cfc

That really is it. The two cache attributes tell ColdFusion to use secondary cache for this data. From now on a call like:

To help work out what secondary cache is doing I did two things in User.cfc:

1. added a version field which would definitively show when Hibernate had updated the record (another very cool underused feature)
2. added my own getter to verify that secondary cache returns data via the cfc and not just flat

Now there are four types of cacheuse, here are the differences:

• transactional cache is updated on the update statement and the next request of the data will get it straight from cache and not run a select
• nonstrict-read-write Cache is updated on the first select after data is updated or inserted
• read-write I could not get this to work with caching (docs say its the worst performing anyway)
• read-only Only works with data that will not be updated. Produces an error if you attempt to update or insert records. (Marc Esher wrote a long, funny blog entry about using secondary cache to cache for a read-only view)

I'm going to look more into secondary cache and hopefully come up with real examples but for now I know two things: 1) its very powerful, 2) its very easy to use.

* If ColdFusion is started from the Builder server panel the console tab will display all the SQL Hibernate produces. I used it to see when Hibernate went to cache and when the database.

Here is a straightforward way to start Solr when running ColdFusion 9 on a Mac. If it is not running in the ColdFusion Administrator under Data & Services > ColdFusion Collections the following message will appear:

Unable to retrieve collections from any of the Search Services. Please verify that atleast one of ColdFusion Search Services is installed and running.

To get it running run the following two lines in Terminal:

Thats for multi-instance installs for ColdFusion 9 standard install it will be:

Within seconds Solr will be up and running and collections will show on the page above.

(As a quick note don't delete core0 collection as its needed to get everything up and running.)

Running multiple ColdFusion instances is pretty awesome from failover to performance. Sharing data among them throws up a couple possibilities:

• Use a database (save from one, retrieve from another)
• Use ColdFusion 9.0.1 with Distributed Ehcache and then use the cachePut() and cacheGet() functions.
• Use objectSave() and objectLoad() while saving the object to disk.

Recently I needed to do just this with a dataset of structures and arrays that could have changed at anytime. This made option 1 a little tricky. Option 2 was my favorite but required some extra installs with Ehcache (which is a simple 7 step process, but would have required testing, yada, yada). So, option 3 was what we picked and it worked well.

Here is a simple example of how to do it. First create and save your object:

The objectSave() function takes two arguments; first the object, and second the filename. By default it saves it relatively to the running file. The documentation use a ".out" extension but I think a ".cfo" is more fun (ColdFusion Object or if you need to please your CFO tell them you named it after them!).

Retrieval is just as easy:

And that will work across any instance.

If you have large objects you may run into issues with writes and saves coming at the same time. In that case wrap the objectLoad() in a try/catch.

Data types for dates and times are slightly different across databases which often makes life difficult for us developers. ColdFusion ORM makes this pretty simple:

• For a date with time use ormtype="timestamp"
• Just a date use ormtype="date"

There are no other options!

Full example with the property tag:

ColdFusion 9.0.1 introduces a small bug into cfdump where changes to function arguments will not show up in cfdump if the dump has already been run once.  I don't know the background but my guess is that in order to speed things up CF is using the cache to store the dump of function code.

The fix to this is to clear from cache anything that starts with _cf_dump with this code:

<cfscript> for ( key in cacheGetAllIds() ) {  if ( left( key, 8) == "_CF_DUMP") {   cacheRemove( key );  } }  </cfscript>

If you are doing a lot of looking at cfdump after changing methods (something I almost never do) then you can place that code in Application.cfc onRequestEnd.

ColdFusion 9.0.1 brings multiple datasources support for ORM. Here is how to set it up.

In this example, we are going to use both the cfartgallery and cfbookclub datasources.  In the Application.cfc we will make the cfartgallery the default datasource.  Optionally we are going to set the dbcreate rules differently for the two datasources.  (We could also do this for the schema, catalog, dialect and sqlscript settings.)

From cfartgallery we will use the Art table as an object.  As cfartgallery is the default datasource, this is set up as usual:

For any tables from the cfbookclub datasource there is an additional attribute "datasource" for the component:

We can then get the data and dump it like so:

 Pretty straighforward and pretty cool stuff.