I'm considering using Hibernate but I'm a bit confused by something. The documentation all assumes I know what Hibernate is and how it's designed. This is a real blind spot in your documentation.

It looks like Hibernate binaries are included with an application I develop. Why is nHibernate code included in my application?

The database schema is fixed and will not be changed at run time. Shouldn't it generate source code which I then compile? Is it loading, parsing the xml object to database mapping, and generating queries dynamically each time my application runs? This would be horribly inefficient since this only needs to be done once and will never change.

The docs speak about caching data and lazy writes. Is this going to work in a multiuser database?

Hello!

I might not be perfect to answer some of your questions, but that seldomly stops me :)

You should not have to care how NHibernate is internally designed, you just should know what it is for and how you use it. Generally it is "only" for mapping relational data to object instances in memory. With tools its also able to automatically update your db-schema or even generate code.

Technically it is just a tool to cut down development time. Its as always a trade between comfy programming and performance. So if you have lots of time and costs dont matter, you are more performant with direct queries on stored procedures. But consider, generating the queries themselfes is not the bottelneck of this system. Transforming the results to object instances is much more time consuming.

I for myselfe like it very much, to just edit some mapping document together with the classes and have the structure automatically updated with one line of code. As for multiuser environments, this also depends on the DB youre going to use. I developed my Business Layer in a way, that the cache is flushed on every save. So the biggest multiuser issue is anyways allways recognizing changes of other users.

Greetings!
Zorgoban

Ugh. Sorry to be disagreeable, but this is wrong on many levels. I really care about the quality of what I produce. I don't just slap stuff together without understanding how it works and picking the right tool for the job.



I don't think the costs are terribly high to write sql statements. Even if it were high I could probably find a code generator or use serialization in .NET.

Thanks for taking out the time to help. :)

I understand your quality standard, but when it comes to using a 3rd party product, you naturally end up in trusting the quality of the product when it meets all your requirements without knowing each single line of sourcecode.



NHibernate is not a sql-generator, nor is a sql-generator equivalent to what NHibernate does. Its about getting data out of the db into objects that will contain logic to work with in the bussines layer. you dont get that without effort when working with data sets.

So how is what nHibernate does different than writing code to persist objects and a business layer to contain the application logic?

My problem is seeing why I need hibernate code at RUN time. I don't change databases, the database structure doesn't change, and my objects that model it don't change. Why isn't this just static code? What's the advantage I gain by having Hibernate code in exchange for the extra performance I pay for running that code?

It seems like this tool is something that makes the development easier at the expense of application size and performance.