hi,
Kind of new to nHibernate.
I find a bit strange and not good for performance that in order to delete a record into database you have to retrieve the entity and then delete.2 network trips.

Can't we just delete based on the id ?
may be I am missing the obvious.

Can somebody help/explain?
Thanks

Yes, you could use an HQL query, something like:
"delete from myClass where myClass.ID = " + myClassId;

It will still require 2 trips. When you delete from NHibernate, then entities deleted will first be loaded into the cache. So, if you said something like:
"delete from myClass"
All your myClass instances in the database would have to be loaded into the cache first and then the delete would happen second.

Two calls for a delete isn't too bad, you are still using the same connection. If, however, you have alot of deletes, you can enlist ADOCommands into the current NHibernate transaction.
See - http://www.lostechies.com/blogs/joshua_ ... ction.aspx

This is the type of approach I take for batch operations. Just remember, when you enlist a command, the operations in the command are unmanaged and can screw up your session state. I think you'd have to call session.Refresh(entity) if a loaded entity was dependent on the command executed (such as deleted children). It would probably be safest to just call session.Flush() before executing the enlisted command and session.Clear() after.

How that helps.

Thank you very much for your reply.
However i think this is extremely bad for performance.
2 TRIPS TO MAKE A DELETE ???.

Also and this is because I am new to Nhibernate.That sometimes if you are not careful how you code it will make an insert instead of an update.
I believe that you have to use the same instantiation of your class otherwise it beleives its a new entity and perform an insert instead of an update.
Have you had it before?

Is it possible to retrieve things from the cache without having a network trip to the database?If so where can I find an example?
Thanks a lot

Again, the entities are managed by NHibernate through the cash, so they must first be loaded then deleted. Enlisting a command for the delete wil bypass this.


Right, if you create a new instance it is a different entity. NHibernate serves as your repository and you use it to manage persitable entities. If you have a new instance and the ID is not set yet, it's new and saving it will add it to the DB. You can put unique constraints in your mapping if you have properties that are unique given a particular class (without having to add constraint in database itself). I can't remember when I've accidentally inserted something last, but I've been using NHibernate fro about 3+ years now. Just remember, an entity is identified by it's reference, not it's attributes. If there's no ID, it's new...if you pulled it from the repository, it's a pre-existing object.


Most definitely...in fact this is one of the great strengths of NHibernate. Once you insert an item or retrieve an item from the repository it is in the cash by default (unless you evict the object or clear the session). You should read about the difference of Get and Load. Also, you can set the session's FlushMode so that data is only flushed to the database on Commits if you like. Above all I'd recommend handling all saves, deletes, etc. in transactions and deal with well defined units of work.
Once you get the hang of it, you'll see that you can often reduce round trips thanks to the cache. I'd recommend setting up a log for NHibernate using log4net and setting it to debug. You can see exactly what's going on on the back end.

Fantastic replies.Thanks a lot

I guess I have to learn how to use the cache properly and understand the difference between get and load.I suppose one of the 2 doensnt make a trip to the db.

The problem is that the community of NHibernate is getting bigger but there are no books about it,where you can quickly learn these vital bits.
Or some good examples to download.
Thanks again

and don't forget to rate people's posts if they help you...some folks need the street cred (LOL).

BTW, have you used log4net before? It's been a major help for me with debugging.

I havent used log4dotnet yet.I will.
I will find out how you rate somebody post and do it.thanks

In a nutshell here's how you do it. If you have any questions I'd be glad to help. I can't live without it and alot of times people will ask for logs when you seek help on the forums.

First off, reference (I do so by browse) log4net.dll, a copy is in the NHibernate bin.

In your code (say, at the beginning of a unit test, on load of a page, at the beginning of main, whatever) you'll call the log4net configurator

using log4net.Config;
...
XmlConfigurator.Configure();

You'll also need to set up your log4net configuration and NHibernate logger in either your web.config or app.config like this:



I love log4net and use it alot in my own code, not just with NHibernate.

More info on configurations are here:
http://logging.apache.org/log4net/relea ... ation.html
And appenders here:
http://logging.apache.org/log4net/relea ... mples.html
Quick and dirty guide:
http://haacked.com/archive/2005/03/07/C ... tions.aspx