Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Clinton Begin]

Generally speaking, ObjectFactory is reponsible for instantiating
classes (including providing constructor parameters).  The
MetaClass/MetaObject library, including the ObjectWrapper abstraction
are reponsible for manipulating / mutating existing objects.

Both ObjectWrapper and ObjectFactory can be overridden.  Can you use a
combination of the two to do what you need?   There are examples of
both in the iBATIS source.

Also, you should be able to find the scala example/plug-in if you
search the mailing list archives.  I'll post it here when I'm at a
computer, if you haven't already found it.

Cheers,
Clinton

On 2010-02-24, Alexei Sokolov  wrote:
> Clinton,
>
> Sorry, I did not mean to criticize the framework. I'm just wondering if I
> understand how you do things, and if the code can be improved upon.
>
> What I see so far is that I cannot separate object construction mechanism
> from the client view of the object. Think StringBuilder/String in java,
> ListBuffer/List in scala, and other similar classes. I want users of my
> classes to see String, List, etc., but I want ibatis to use StringBuilder
> (stringbuilder is not the real usecase, but just to show my point) and
> ListBuffer to construct them. It could be due to how I want users to see the
> data, or it could be due to the fact that I can construct the object more
> efficiently this way, or I want it to be immutable.
>
> If you'll say that such pattern is outside of ibatis scope, it is OK.
> However, I think this refactoring may benefit others, including scala
> users.
>
> Alex
>
> On Wed, Feb 24, 2010 at 8:44 PM, Clinton Begin
> wrote:
>
>> LOL... sorry I thought that was funny.  :-)
>>
>> But really, you're really polarizing this discussion with broad criticisms
>> of how our software is designed.  It's hard to comment on your needs or to
>> help you.
>>
>> I'm not exactly sure what you're trying to do, but you're digging pretty
>> deep into the guts of the system, parts that weren't really intended for
>> other people to extend or change (or at the very least at your own risk).
>>
>> So it seems to me that if you have to dig that deep into the core of the
>> library, it's probably the wrong fit and you'll have an uphill battle and
>> a
>> poor experience overall.
>>
>> So rather than criticize our framework because it doesn't do what you need
>> it to, it might be best if you look for a better fit.  There are others.
>> Hibernate is one, and yes it's well designed.
>>
>> As for a friendly community, I can assure you that if you broadly
>> criticize
>> them because their square peg doesn't fit into the round hole, you'll
>> probably meet with a similar reaction.
>>
>> Cheers,
>> Clinton
>>
>>
>> On Wed, Feb 24, 2010 at 9:03 PM, Alexei Sokolov
>> wrote:
>>
>>> Not sure what are you trying to say.
>>>
>>> That hibernate has friendly community? Or that they have a better
>>> designed
>>> library?
>>>
>>>
>>> -- Alex
>>> Sent from my mobile
>>>
>>> On Feb 24, 2010, at 7:28 PM, Clinton Begin 
>>> wrote:
>>>
>>>  Yes, iBATIS has a mechanism that allows you to do what you want to do.
>>> It's called Hibernate.
>>>
>>> https://www.hibernate.org/
>>>
>>> Enjoy.
>>>
>>> On Wed, Feb 24, 2010 at 8:11 PM, Alexei Sokolov
>>> <
>>> alexei.soko...@gmail.com> wrote:
>>>
 Well, Microsoft Office has Excel, which is a good spreadsheet. Does
 ibatis has a mechanism that allows me to do what I want to do?

 If intented use for objectwrapper was scala, why didn't you call it
 scalaobjectwrapper? Obviously you had other potential uses in mind.

 BTW, can you point me to some scala examples? Can ibatis map java lists
 and maps to scala lists and maps?

 Thanks,
 -- Alex
 Sent from my mobile

 On Feb 24, 2010, at 6:35 PM, Clinton Begin < 
 clinton.be...@gmail.com> wrote:

 What you're trying to do with ObjectWrapper is beyond its intended use.
 It was never a consideration nor an intention for the ObjectWrapper to
 instantiate objects.  ObjectWrapper was originally implemented to
 support
 Scala types, and perhaps to support a custom property naming convention
 or
 some other type (e.g. XML).

 It was not intended to change or intercept the result object lifecycle
 or
 override the result object entirely.

 You can call it a design problem if you like.  But it's kind of like
 saying Microsoft Word has a design problem because it isn't a very good
 spreadsheet.

 Clinton

 On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov
 <
 alexei.soko...@gmail.com> wrote:

> Hello everyone,
>
> I'm new here, but either I don't understand ibatis internals or you
> have
> a design problem. First, let me share my understanding of how ibatis
> handles
> mapping of result sets to beans:
>
> In *ResultSetHandler.getRowValue() method you create object instance
> using objectfac

Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Alexei Sokolov]

Clinton,

Sorry, I did not mean to criticize the framework. I'm just wondering if I
understand how you do things, and if the code can be improved upon.

What I see so far is that I cannot separate object construction mechanism
from the client view of the object. Think StringBuilder/String in java,
ListBuffer/List in scala, and other similar classes. I want users of my
classes to see String, List, etc., but I want ibatis to use StringBuilder
(stringbuilder is not the real usecase, but just to show my point) and
ListBuffer to construct them. It could be due to how I want users to see the
data, or it could be due to the fact that I can construct the object more
efficiently this way, or I want it to be immutable.

If you'll say that such pattern is outside of ibatis scope, it is OK.
However, I think this refactoring may benefit others, including scala
users.

Alex

On Wed, Feb 24, 2010 at 8:44 PM, Clinton Begin wrote:

> LOL... sorry I thought that was funny.  :-)
>
> But really, you're really polarizing this discussion with broad criticisms
> of how our software is designed.  It's hard to comment on your needs or to
> help you.
>
> I'm not exactly sure what you're trying to do, but you're digging pretty
> deep into the guts of the system, parts that weren't really intended for
> other people to extend or change (or at the very least at your own risk).
>
> So it seems to me that if you have to dig that deep into the core of the
> library, it's probably the wrong fit and you'll have an uphill battle and a
> poor experience overall.
>
> So rather than criticize our framework because it doesn't do what you need
> it to, it might be best if you look for a better fit.  There are others.
> Hibernate is one, and yes it's well designed.
>
> As for a friendly community, I can assure you that if you broadly criticize
> them because their square peg doesn't fit into the round hole, you'll
> probably meet with a similar reaction.
>
> Cheers,
> Clinton
>
>
> On Wed, Feb 24, 2010 at 9:03 PM, Alexei Sokolov 
> wrote:
>
>> Not sure what are you trying to say.
>>
>> That hibernate has friendly community? Or that they have a better designed
>> library?
>>
>>
>> -- Alex
>> Sent from my mobile
>>
>> On Feb 24, 2010, at 7:28 PM, Clinton Begin 
>> wrote:
>>
>>  Yes, iBATIS has a mechanism that allows you to do what you want to do.
>> It's called Hibernate.
>>
>> https://www.hibernate.org/
>>
>> Enjoy.
>>
>> On Wed, Feb 24, 2010 at 8:11 PM, Alexei Sokolov <
>> alexei.soko...@gmail.com> wrote:
>>
>>> Well, Microsoft Office has Excel, which is a good spreadsheet. Does
>>> ibatis has a mechanism that allows me to do what I want to do?
>>>
>>> If intented use for objectwrapper was scala, why didn't you call it
>>> scalaobjectwrapper? Obviously you had other potential uses in mind.
>>>
>>> BTW, can you point me to some scala examples? Can ibatis map java lists
>>> and maps to scala lists and maps?
>>>
>>> Thanks,
>>> -- Alex
>>> Sent from my mobile
>>>
>>> On Feb 24, 2010, at 6:35 PM, Clinton Begin < 
>>> clinton.be...@gmail.com> wrote:
>>>
>>> What you're trying to do with ObjectWrapper is beyond its intended use.
>>> It was never a consideration nor an intention for the ObjectWrapper to
>>> instantiate objects.  ObjectWrapper was originally implemented to support
>>> Scala types, and perhaps to support a custom property naming convention or
>>> some other type (e.g. XML).
>>>
>>> It was not intended to change or intercept the result object lifecycle or
>>> override the result object entirely.
>>>
>>> You can call it a design problem if you like.  But it's kind of like
>>> saying Microsoft Word has a design problem because it isn't a very good
>>> spreadsheet.
>>>
>>> Clinton
>>>
>>> On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov 
>>> <
>>> alexei.soko...@gmail.com> wrote:
>>>
 Hello everyone,

 I'm new here, but either I don't understand ibatis internals or you have
 a design problem. First, let me share my understanding of how ibatis 
 handles
 mapping of result sets to beans:

 In *ResultSetHandler.getRowValue() method you create object instance
 using objectfactory. then you create metaobject for it, which in turn
 creates objectwrapper for that object instance. Then you use meta object
 methods to populate property values on that object instance. During this
 process both metaobject and objectwrapper essentially write values directly
 to the object instance returned from object factory.

 Now, consider this scenario: let's say my object wrapper constructs
 intermediate object first and populates its values. Than when I'm done
 processing a row, I want to create actual object instance that will be
 returned to the client. How do I do this with the current implementation?
 Ideally, it will be nice if ObjectWrapper had a method, say
 'getPopulatedInstance()' and then line 179 in
 FastResultSetHandler.getRowValue() (and oth

Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Clinton Begin]

LOL... sorry I thought that was funny.  :-)

But really, you're really polarizing this discussion with broad criticisms
of how our software is designed.  It's hard to comment on your needs or to
help you.

I'm not exactly sure what you're trying to do, but you're digging pretty
deep into the guts of the system, parts that weren't really intended for
other people to extend or change (or at the very least at your own risk).

So it seems to me that if you have to dig that deep into the core of the
library, it's probably the wrong fit and you'll have an uphill battle and a
poor experience overall.

So rather than criticize our framework because it doesn't do what you need
it to, it might be best if you look for a better fit.  There are others.
Hibernate is one, and yes it's well designed.

As for a friendly community, I can assure you that if you broadly criticize
them because their square peg doesn't fit into the round hole, you'll
probably meet with a similar reaction.

Cheers,
Clinton

On Wed, Feb 24, 2010 at 9:03 PM, Alexei Sokolov wrote:

> Not sure what are you trying to say.
>
> That hibernate has friendly community? Or that they have a better designed
> library?
>
>
> -- Alex
> Sent from my mobile
>
> On Feb 24, 2010, at 7:28 PM, Clinton Begin 
> wrote:
>
> Yes, iBATIS has a mechanism that allows you to do what you want to do.
> It's called Hibernate.
>
> https://www.hibernate.org/
>
> Enjoy.
>
> On Wed, Feb 24, 2010 at 8:11 PM, Alexei Sokolov <
> alexei.soko...@gmail.com> wrote:
>
>> Well, Microsoft Office has Excel, which is a good spreadsheet. Does ibatis
>> has a mechanism that allows me to do what I want to do?
>>
>> If intented use for objectwrapper was scala, why didn't you call it
>> scalaobjectwrapper? Obviously you had other potential uses in mind.
>>
>> BTW, can you point me to some scala examples? Can ibatis map java lists
>> and maps to scala lists and maps?
>>
>> Thanks,
>> -- Alex
>> Sent from my mobile
>>
>> On Feb 24, 2010, at 6:35 PM, Clinton Begin < 
>> clinton.be...@gmail.com> wrote:
>>
>> What you're trying to do with ObjectWrapper is beyond its intended use. It
>> was never a consideration nor an intention for the ObjectWrapper to
>> instantiate objects.  ObjectWrapper was originally implemented to support
>> Scala types, and perhaps to support a custom property naming convention or
>> some other type (e.g. XML).
>>
>> It was not intended to change or intercept the result object lifecycle or
>> override the result object entirely.
>>
>> You can call it a design problem if you like.  But it's kind of like
>> saying Microsoft Word has a design problem because it isn't a very good
>> spreadsheet.
>>
>> Clinton
>>
>> On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov 
>> <
>> alexei.soko...@gmail.com> wrote:
>>
>>> Hello everyone,
>>>
>>> I'm new here, but either I don't understand ibatis internals or you have
>>> a design problem. First, let me share my understanding of how ibatis handles
>>> mapping of result sets to beans:
>>>
>>> In *ResultSetHandler.getRowValue() method you create object instance
>>> using objectfactory. then you create metaobject for it, which in turn
>>> creates objectwrapper for that object instance. Then you use meta object
>>> methods to populate property values on that object instance. During this
>>> process both metaobject and objectwrapper essentially write values directly
>>> to the object instance returned from object factory.
>>>
>>> Now, consider this scenario: let's say my object wrapper constructs
>>> intermediate object first and populates its values. Than when I'm done
>>> processing a row, I want to create actual object instance that will be
>>> returned to the client. How do I do this with the current implementation?
>>> Ideally, it will be nice if ObjectWrapper had a method, say
>>> 'getPopulatedInstance()' and then line 179 in
>>> FastResultSetHandler.getRowValue() (and other such places) could be
>>> rewritten like
>>>
>>> resultObject = foundValues ?
>>> metaObject.getObjectWrapper().getPopulatedInstance() : null;
>>>
>>> instead of
>>>
>>> resultObject = foundValues ? resultObject : null;
>>>
>>> In other words, it would be nice if objectwrapper is able to create new
>>> instances in addition to being able to work as decorator.
>>>
>>> Does it make sense at all?
>>>
>>> Thanks,
>>> Alex
>>>
>>>
>>
>

Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Alexei Sokolov]

Not sure what are you trying to say.

That hibernate has friendly community? Or that they have a better  
designed library?


-- Alex
Sent from my mobile

On Feb 24, 2010, at 7:28 PM, Clinton Begin   
wrote:


Yes, iBATIS has a mechanism that allows you to do what you want to  
do.  It's called Hibernate.


https://www.hibernate.org/

Enjoy.

On Wed, Feb 24, 2010 at 8:11 PM, Alexei Sokolov > wrote:
Well, Microsoft Office has Excel, which is a good spreadsheet. Does  
ibatis has a mechanism that allows me to do what I want to do?


If intented use for objectwrapper was scala, why didn't you call it  
scalaobjectwrapper? Obviously you had other potential uses in mind.


BTW, can you point me to some scala examples? Can ibatis map java  
lists and maps to scala lists and maps?


Thanks,
-- Alex
Sent from my mobile

On Feb 24, 2010, at 6:35 PM, Clinton Begin   
wrote:


What you're trying to do with ObjectWrapper is beyond its intended  
use. It was never a consideration nor an intention for the  
ObjectWrapper to instantiate objects.  ObjectWrapper was originally  
implemented to support Scala types, and perhaps to support a custom  
property naming convention or some other type (e.g. XML).


It was not intended to change or intercept the result object  
lifecycle or override the result object entirely.


You can call it a design problem if you like.  But it's kind of  
like saying Microsoft Word has a design problem because it isn't a  
very good spreadsheet.


Clinton

On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov > wrote:

Hello everyone,

I'm new here, but either I don't understand ibatis internals or you  
have a design problem. First, let me share my understanding of how  
ibatis handles mapping of result sets to beans:


In *ResultSetHandler.getRowValue() method you create object  
instance using objectfactory. then you create metaobject for it,  
which in turn creates objectwrapper for that object instance. Then  
you use meta object methods to populate property values on that  
object instance. During this process both metaobject and  
objectwrapper essentially write values directly to the object  
instance returned from object factory.


Now, consider this scenario: let's say my object wrapper constructs  
intermediate object first and populates its values. Than when I'm  
done processing a row, I want to create actual object instance that  
will be returned to the client. How do I do this with the current  
implementation? Ideally, it will be nice if ObjectWrapper had a  
method, say 'getPopulatedInstance()' and then line 179 in  
FastResultSetHandler.getRowValue() (and other such places) could be  
rewritten like


resultObject = foundValues ? metaObject.getObjectWrapper 
().getPopulatedInstance() : null;


instead of

resultObject = foundValues ? resultObject : null;

In other words, it would be nice if objectwrapper is able to create  
new instances in addition to being able to work as decorator.


Does it make sense at all?

Thanks,
Alex

Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Clinton Begin]

Yes, iBATIS has a mechanism that allows you to do what you want to do.  It's
called Hibernate.

https://www.hibernate.org/

Enjoy.

On Wed, Feb 24, 2010 at 8:11 PM, Alexei Sokolov wrote:

> Well, Microsoft Office has Excel, which is a good spreadsheet. Does ibatis
> has a mechanism that allows me to do what I want to do?
>
> If intented use for objectwrapper was scala, why didn't you call it
> scalaobjectwrapper? Obviously you had other potential uses in mind.
>
> BTW, can you point me to some scala examples? Can ibatis map java lists and
> maps to scala lists and maps?
>
> Thanks,
> -- Alex
> Sent from my mobile
>
> On Feb 24, 2010, at 6:35 PM, Clinton Begin 
> wrote:
>
> What you're trying to do with ObjectWrapper is beyond its intended use. It
> was never a consideration nor an intention for the ObjectWrapper to
> instantiate objects.  ObjectWrapper was originally implemented to support
> Scala types, and perhaps to support a custom property naming convention or
> some other type (e.g. XML).
>
> It was not intended to change or intercept the result object lifecycle or
> override the result object entirely.
>
> You can call it a design problem if you like.  But it's kind of like saying
> Microsoft Word has a design problem because it isn't a very good
> spreadsheet.
>
> Clinton
>
> On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov <
> alexei.soko...@gmail.com> wrote:
>
>> Hello everyone,
>>
>> I'm new here, but either I don't understand ibatis internals or you have a
>> design problem. First, let me share my understanding of how ibatis handles
>> mapping of result sets to beans:
>>
>> In *ResultSetHandler.getRowValue() method you create object instance using
>> objectfactory. then you create metaobject for it, which in turn creates
>> objectwrapper for that object instance. Then you use meta object methods to
>> populate property values on that object instance. During this process both
>> metaobject and objectwrapper essentially write values directly to the object
>> instance returned from object factory.
>>
>> Now, consider this scenario: let's say my object wrapper constructs
>> intermediate object first and populates its values. Than when I'm done
>> processing a row, I want to create actual object instance that will be
>> returned to the client. How do I do this with the current implementation?
>> Ideally, it will be nice if ObjectWrapper had a method, say
>> 'getPopulatedInstance()' and then line 179 in
>> FastResultSetHandler.getRowValue() (and other such places) could be
>> rewritten like
>>
>> resultObject = foundValues ?
>> metaObject.getObjectWrapper().getPopulatedInstance() : null;
>>
>> instead of
>>
>> resultObject = foundValues ? resultObject : null;
>>
>> In other words, it would be nice if objectwrapper is able to create new
>> instances in addition to being able to work as decorator.
>>
>> Does it make sense at all?
>>
>> Thanks,
>> Alex
>>
>>
>

Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Alexei Sokolov]

Well, Microsoft Office has Excel, which is a good spreadsheet. Does  
ibatis has a mechanism that allows me to do what I want to do?


If intented use for objectwrapper was scala, why didn't you call it  
scalaobjectwrapper? Obviously you had other potential uses in mind.


BTW, can you point me to some scala examples? Can ibatis map java  
lists and maps to scala lists and maps?


Thanks,
-- Alex
Sent from my mobile

On Feb 24, 2010, at 6:35 PM, Clinton Begin   
wrote:


What you're trying to do with ObjectWrapper is beyond its intended  
use. It was never a consideration nor an intention for the  
ObjectWrapper to instantiate objects.  ObjectWrapper was originally  
implemented to support Scala types, and perhaps to support a custom  
property naming convention or some other type (e.g. XML).


It was not intended to change or intercept the result object  
lifecycle or override the result object entirely.


You can call it a design problem if you like.  But it's kind of like  
saying Microsoft Word has a design problem because it isn't a very  
good spreadsheet.


Clinton

On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov > wrote:

Hello everyone,

I'm new here, but either I don't understand ibatis internals or you  
have a design problem. First, let me share my understanding of how  
ibatis handles mapping of result sets to beans:


In *ResultSetHandler.getRowValue() method you create object instance  
using objectfactory. then you create metaobject for it, which in  
turn creates objectwrapper for that object instance. Then you use  
meta object methods to populate property values on that object  
instance. During this process both metaobject and objectwrapper  
essentially write values directly to the object instance returned  
from object factory.


Now, consider this scenario: let's say my object wrapper constructs  
intermediate object first and populates its values. Than when I'm  
done processing a row, I want to create actual object instance that  
will be returned to the client. How do I do this with the current  
implementation? Ideally, it will be nice if ObjectWrapper had a  
method, say 'getPopulatedInstance()' and then line 179 in  
FastResultSetHandler.getRowValue() (and other such places) could be  
rewritten like


resultObject = foundValues ? metaObject.getObjectWrapper 
().getPopulatedInstance() : null;


instead of

resultObject = foundValues ? resultObject : null;

In other words, it would be nice if objectwrapper is able to create  
new instances in addition to being able to work as decorator.


Does it make sense at all?

Thanks,
Alex

Re: help with ibatis 3 & objectwrapper, metaobject, etc.

[Clinton Begin]

What you're trying to do with ObjectWrapper is beyond its intended use. It
was never a consideration nor an intention for the ObjectWrapper to
instantiate objects.  ObjectWrapper was originally implemented to support
Scala types, and perhaps to support a custom property naming convention or
some other type (e.g. XML).

It was not intended to change or intercept the result object lifecycle or
override the result object entirely.

You can call it a design problem if you like.  But it's kind of like saying
Microsoft Word has a design problem because it isn't a very good
spreadsheet.

Clinton

On Wed, Feb 24, 2010 at 11:11 AM, Alexei Sokolov
wrote:

> Hello everyone,
>
> I'm new here, but either I don't understand ibatis internals or you have a
> design problem. First, let me share my understanding of how ibatis handles
> mapping of result sets to beans:
>
> In *ResultSetHandler.getRowValue() method you create object instance using
> objectfactory. then you create metaobject for it, which in turn creates
> objectwrapper for that object instance. Then you use meta object methods to
> populate property values on that object instance. During this process both
> metaobject and objectwrapper essentially write values directly to the object
> instance returned from object factory.
>
> Now, consider this scenario: let's say my object wrapper constructs
> intermediate object first and populates its values. Than when I'm done
> processing a row, I want to create actual object instance that will be
> returned to the client. How do I do this with the current implementation?
> Ideally, it will be nice if ObjectWrapper had a method, say
> 'getPopulatedInstance()' and then line 179 in
> FastResultSetHandler.getRowValue() (and other such places) could be
> rewritten like
>
> resultObject = foundValues ?
> metaObject.getObjectWrapper().getPopulatedInstance() : null;
>
> instead of
>
> resultObject = foundValues ? resultObject : null;
>
> In other words, it would be nice if objectwrapper is able to create new
> instances in addition to being able to work as decorator.
>
> Does it make sense at all?
>
> Thanks,
> Alex
>
>