Fabio, I'm sorry that you think that way, I sincerely do think there is a bug in NHibernate. I am not asking for expert advise, otherwise, I would have gone for the NHUsers mailing list, like I usually do... You always ask people to listen to what you say. I believe you are not listening to what I am saying and you are discarding the chance that there is indeed a bug. Unless you want me to continue, this will be my last attempt in convincing you, I will drop this thread. The course of events is:
1) if I load an entity from its id, when I check is dirty, no saveorupdate event is fired (I have a listener other than the default); 2) If I load another entity that has a lazy property to the "problematic" entity, when I check is dirty, it fires the saveorupdate event (which, eventually, will change it, but it was unchanged when the saveorupdate event was raised); 3) the two entities are in memory, both are unchanged, I am using the code you published in your blog for checking it; why is the saveorupdate event fired? It will cause an unchanged entity to become dirty! 4) The Equals/GetHashCode implementation is the same on both tests. Guys: can anyone please help me understand this? The issue is http://216.121.112.228/browse/NH-2727. Fabio: you have always been a great help, and I sincerely thank you for your time! RP On May 24, 3:36 pm, Fabio Maulo <[email protected]> wrote: > There was and are some cases where ppl file a bug to have an answer by some > expert. > There cases where the expert may fall into the trap. > Have a look to your equality comparer and always check the state of the > entity (using "your" extension-method) before change the state of the > entity. > > > > > > > > > > On Tue, May 24, 2011 at 11:25 AM, Ricardo Peres <[email protected]> wrote: > > Fabio, > > > Forgive me, but I still don't understand why, in one occasion, it > > fires the event, and not on another, depending on how the entity was > > loaded. I thought all entities were equal, but it seems some are more > > equal than others! ;-) > > Please, if you have some time, do check my code; it is very simple and > > straightforward. > > > RP > > > On May 24, 3:20 pm, Fabio Maulo <[email protected]> wrote: > > > NH does not check a state of an loaded entity through > > INotifyPropertyChanged > > > so the only way to know if a loaded entity-state is dirty is asking to > > those > > > events if they have something to do. > > > A similar behavior is done by the flush-mode=auto when you fire a > > query... > > > In practice: > > > if you are using lazy-properties (a simple value with lazy or a relation > > > with no-proxy) we may have an issue related to it... but this is only a > > > guess just because only you and God knows your mappings/classes. > > > > The fact that session.IsDirty fire the SaveOrUpdate event, where neither > > > inserts nor deletes are presents, is not an issue, instead it is the > > default > > > behavior that you can completely override. > > > > On Tue, May 24, 2011 at 11:11 AM, Ricardo Peres <[email protected]> > > wrote: > > > > Fabio, > > > > > Don't get me wrong: I have followed the stack trace, and I know why > > > > this is happening (in the code); I just don't understand it. > > > > First of all: from a conceptual point of view, should the > > > > ISession.IsDirty() fire SaveOrUpdate, on non-dirty entities, or on any > > > > entities at all? > > > > Second: why, if we are loading the entity by its it, the event does > > > > not fire, and if we load it from a property of another entity, it > > > > does? > > > > IMHO, if you answer yes to the first question, there is a bug: it > > > > isn't being fired if the entity is not directly loaded. > > > > I don't want to take your time, just to understand this. Am I the only > > > > one who doesn't understand this behavior? > > > > > Thanks! > > > > > RP > > > > > On May 24, 3:05 pm, Fabio Maulo <[email protected]> wrote: > > > > > The only thing done by IsDirty is just fire an event > > > > > DirtyCheckEvent dcEvent = new DirtyCheckEvent(this); > > > > > IDirtyCheckEventListener[] dirtyCheckEventListener = > > > > > listeners.DirtyCheckEventListeners; > > > > > for (int i = 0; i < dirtyCheckEventListener.Length; i++) > > > > > { > > > > > dirtyCheckEventListener[i].OnDirtyCheck(dcEvent);} > > > > > > return dcEvent.Dirty; > > > > > You can disable/replace/override that event. > > > > > > On Tue, May 24, 2011 at 10:21 AM, Ricardo Peres <[email protected]> > > > > wrote: > > > > > > Fabio, > > > > > > You have closed JIRA issues NH-2727 saying that it is not an issue. > > > > > > Perhaps you can explain me, because this is bugging me, why does > > the > > > > > > following line raise the SaveOrUpdate event and the next doesn't: > > > > > > > //raises SaveOrUpdate > > > > > > User u = session.Query<User>().FirstOrDefault(); > > > > > > UserGroup ug = u.UserGroup.First(); > > > > > > > //does not raise > > > > > > UserGroup ug = session.Query<UserGroup>().FirstOrDefault(); > > > > > > User u = ug.User; > > > > > > > By the way, in general, why does ISession.IsDirty() fire any > > events? > > > > > > Shouldn't it just check the current state of entities in memory? > > > > > > > Thank you for your time, once again! > > > > > > > RP > > > > > > -- > > > > > Fabio Maulo > > > > -- > > > Fabio Maulo > > -- > Fabio Maulo
