[morph-user] Re: BeanUtils patch

[Matt S. <sga...@us...>]

Hi Nathan,

I see what you're trying to do.  I am hesitant to base the flag on 
whether a property is transient or not because properties in other 
objects which have properties (Maps, HttpRequest parameters, even 
ResultSet rows) do not have a notion of being transient or intransient.  
I see where you're going with this idea though, and I agree that the 
functionality would be handy to have out-of-the-box.

Instead of basing the copy on whether the property is transient, how 
about we add a copyEmptyProperties boolean flag that indicates whether 
empty properties should be copied from the source to the destination.  
By default the flag would be set to true, as you suggested.  If 
copyEmptyProperties was false, properties with 'empty' values like a 
null value or an empty string or empty collection would not be copied.

I can code this into the next release, or if you'd like to see your name 
in the @author tag for the class and on the Contributors list of the 
Morph website, feel free to submit a patch :)  I use Eclipse, so I'd 
prefer patches in the "Unified" format.

Matt

negge wrote:

>Matt,
>
>Thanks for the quick response.  Let me tell you how I'd like to use this 
>method.  I am currently using Struts and have a set of Actions that let me 
>edit domain objects in a database.  For each domain object, there is a 
>corresponding form bean that contains just accessor methods.  The domain 
>objects are managed by hibernate, including the object relations - e.g., the 
>many-to-many relationship between a User object and an Organization object.
>
>
>
>The Problem
>
>
>When I go to display the page containing the details for a particular User 
>object, I simply load the domain object via a DAO, and then do:
>
>  User user=userDAO.load(userForm.getId());
>  BeanUtils.copyProperties(userForm,user);
>
>Which copies everything over, including the collection that contains the 
>Organizations this user belongs to.
>
>When I bring up a page that allows you to modify the details of this user 
>(first name, last name, email address, etc. but *not* the organizations since 
>that is managed elsewhere) it uses the same userForm object.  However, when I 
>go to submit this form and save the user, the organizations collection is 
>empty (since its not something to be edited).  Hence
>
>  User user=userDAO.load(userForm.getId());
>  BeanUtils.copyProperties(user,userForm);
>  userDAO.save(user);
>
>just copies an empty collection over the existing one in the user object and 
>when I save it I've lost the organizations information.
>
>
>
>The Solution
>
>
>What I'd like to be able to do is specify this collection as transient (which 
>I already do to support the EqualsBuilder, HashCodeBuilder, and 
>ToStringBuilder) .  Then when simply change the code so that when viewing the 
>details of the object, you call:
>
>  User user=userDAO.load(userForm.getId());
>  BeanUtils.copyProperties(userForm,user,true);
>
>but when saving the details of the object you do:
>
>  User user=userDAO.load(userForm.getId());
>  BeanUtils.copyProperties(user,userForm,false);
>  userDAO.save(user);
>
>This would allow me to only have you editing the fields that make sense (those 
>which are not relationships).
>
>
>I hope this makes sense to you.  I am not familiar enough with the morph 
>library to have presented the problem using your API, please excuse me.  I 
>like your idea of implementing it as a subclass, but strongly believe this is 
>something you should support out of the box.  As in, the default 
>implementation of copy(A,B) should call copy(A,B,true).
>
>I'd be happy to help out with implementing this in any way I can.  I submitted 
>a bug to the BeanUtils issue tracker for the same feature and doubt they will 
>ever get to it.
>
>
>Nathan
>
>
>  
>
>>===== Original Message From Matt Sgarlata <sga...@us...> 
>>    
>>
>=====
>  
>
>>Hi Nathan,
>>
>>Thanks for evaluating Morph!
>>
>>A boolean flag in the PropertyNameMatchingCopier indicating whether
>>transiesnt properties should be copied is an interesting idea.  Would
>>you mind describing how you would use such a property?
>>
>>Morph does not distinguish transient from non-transient properties or
>>static from non-static properties.  The main reason for this is that
>>these notions don't make sense when you start treating other objects
>>such as Maps and HTTP request parameters the same way as you treat
>>Objects.  Maps and HTTP request parameters don't have any notion of
>>transient or static properties.
>>
>>I think the easiest way to accomodate your requirement would be to add a
>>getPropertiesToCopy(Object destination, Object source) method to the
>>PropertyNameMatchingCopier that by default delegated to
>>getPropertiesToCopy().  That way, in a subclass you could override the
>>getPropertiesToCopy method and scan the source object to determine which
>>properties are transient.  Alternatively, if you're only dealing with a
>>couple transformations you could just specify the propertiesToCopy
>>statically by calling the setPropertiesToCopy method.  I will definitely
>>be adding the getPropertiesToCopy(Object destination, Object source)
>>method in the next version of Morph.  If you like, I can do a release
>>for you soon that includes this method.
>>
>>Matt
>>
>>negge wrote:
>>
>>    
>>
>>>Matt,
>>>
>>>I was just experiencing some frustration with BeanUtils and thought I'd take 
>>>      
>>>
>a
>  
>
>>>look at the Morph package.  I was hoping the PropertyNameMatchingCopier had
>>>the ability to allow you to specify a boolean indicating whether or not it
>>>should copy member variables that are marked as transient.  This 
>>>      
>>>
>functionality
>  
>
>>>exists in the EqualsBuilder, HashCodeBuilder, ToStringBuilder so you can 
>>>      
>>>
>mark
>  
>
>>>primary keys or Collection objects as transient and they will not be used in
>>>computing the hashcode or equality of two objects.
>>>
>>>I'd hope to see it implemented so I could just execute
>>>
>>> copy(user,userForm,false);
>>>
>>>and have it work.
>>>
>>>Thanks for creating a very useful looking library.
>>>
>>>Nathan
>>>
>>>
>>>
>>>      
>>>
>>>>===== Original Message From Matt Sgarlata <sga...@us...>
>>>>
>>>>
>>>>        
>>>>
>>>=====
>>>
>>>
>>>      
>>>
>>>>Hi Nathan,
>>>>
>>>>I noticed the patch you submitted to BeanUtils.  If you're anything like
>>>>me, you're starting to reach the point of frustration with the package.
>>>>I discovered that as long as I did exactly what BeanUtils was designed
>>>>to do, I didn't run into any problems.  However, as soon as I tried to
>>>>customize things for my application I ran into serious problems.
>>>>
>>>>This led me to create a similar framework that is much more configurable
>>>>and flexible than BeanUtils.  I encourage you to check it out at
>>>>morph.sourceforge.net.  It is in late beta and already offers pretty
>>>>much all of the features of BeanUtils.  Hopefully it can help you solve
>>>>some of your problems!
>>>>
>>>>Matt
>>>>
>>>>
>>>>        
>>>>
>>>
>>>
>>>
>>>      
>>>
>
>
>
>  
>