Feedster found this post for me: Learning NHibernate « Opera Lover Blog. I’m curious how he will fare with real world applications. Hope to hear (or read) more on his progress/findings. I see a lot of my own experience in his post.

This comes from ASP.NET 2.0 Website Programming—Problem – Design – Solution by Marco Bellinaso (ISBN-13: 978-0-7645-8464-0)

Quote: 

“Some people have developed tools called ORM (Object Relational Mappers) utilities, which can make it easier to map database tables, views, and other relational objects to an OOP class. They can dynamically explore the database’s schema and create the wrapper custom entity and domain object classes for you. Some of them work at design time to automatically produce C# classes you can compile into your application, and others work at runtime to perform this mapping function on-the-fly. But sometimes they don’t work as well as we’d like them to, and there can be a considerable learning curve to set up these tools to work with your particular database and usage requirements. Some of the most popular tools for .NET developers are NHibernate (nhibernate.sourceforge.net [open source]), EntityBroker (www.thona-consulting.com/content/products/entitybroker.aspx [free community version and commercial versions available]), and LLBLGen Pro (www.llblgen.com[commercial]). A simpler tool for small to mid-size systems is Paul Wilson’s ORMapper (www.ormapper.net). I decided not to use these tools in this book because I generally prefer a custom-designed mapping solution specific to each application, but I can see that these tools may be your best alternative in some cases. If you have been contracted to develop a new system that will later be maintained by corporate developers, you’ll have to get their buy-in before selecting any third-party tools because some companies have policies that prohibit the use of “unapproved” tools they haven’t officially sanctioned.”

Well,well, strange this comes to my attention exactly now, when I’ve just written two posts on this matter…

I said in a previous post, commenting someone who’s love with NHibernate was over that it is benificial in later stages of an app’s lifecycle to use a framework. I also said it enforces a certain structure on the developer. This led me to reavaluate what I’ve been gaining for using other people’s frameworks.

More and more I start to wonder if the poster of the original isn’t right after all… 

Today there are quite a few frameworks like rails, castle project’s monorail, nhibernate, WilsonORMaper, you name it.

These all promise to simplify .Net development. 

The Castle Project’s homepage says: “…Castle helps you get more done with less code.”
WilsonOR’s about page says:”Why use an O/R Mapper? Most reasons are centered around flexibility and maintainability…”

That’s all true and right of course. But as huenemeca said the learnig curve involved in these is steep.

Personally, I tried the several “getting started” tutorials on their respective websites. After completing the examples you go:”Fantastic!” I mean every one of them “has something”.

The next step sometimes is thinking about a small piece of software you’ve done and wanting to port that idea to the new framework. Well that’s how I do it mostly, it saves me the trouble designing an app. I just pick up the design documentation and I’m off.

Almost as soon as you start, you run into trouble. This won’t work, how do I do this kind of association, and then it’s toiling trough the (mostly) reference documentation.

Reference documentation is in my opinion not a good source for learning. As a beginner you do not see the connection between the different objects, methods and properties. Let alone you see best practices from reference docs. There is no glue. I believe we all can learn much more of complete, well documented apps, examples that have the majority of what is needed to create real world apps.

So I wonder sometimes: “Is it all worth the extra time spent figuring out this new technology?”

It’s inherent to Agile Development that requirements change during the lifespan of an app. Several times I’ve come to the conclusion that I should have used this or that technology instead of the one I’ve chosen. So again I start to wonder if I shouldn’t have created a specific architecture for the app at hand. One you know trough and trough, that is adaptable without the risk of irreversibly breaking something else…

I came across this post here while searching for ramblings on nhibernate:The Honeymoon is Over, My Relationship with NHibernate « Regular Programming

Well, I feel the same sometimes, but I hang in there. The benefits far outweigh the cons once the app is live.