| model objects changing their identity [message #208403] |
Thu, 02 February 2006 09:38  |
Eclipse User |
|
|
|
Originally posted by: christian.sell.netcologne.de
Hello all,
in our GEF applications the model objects override the equals() and
hashCode() methods to compute them based on certain object attributes.
GEF maintains an EditPart registry which uses these objects as keys,
which becomes a problem if the identity attributes of the model objects
change.
My question: is there any recommended practice for cases like the one
above?
thanks,
Christian
|
|
|
|
| Re: model objects changing their identity [message #208464 is a reply to message #208456] |
Thu, 02 February 2006 22:30  |
| Eclipse User |
|
|
|
Originally posted by: christian.sell.netcologne.de
thats what I concluded, too. I guess the real lesson is that the model
should not override the concept of equality/identity at all.
christian
Steven Shaw wrote:
> I would expect that you would have to remove and then re-add the model
> objects from the registry under these conditions.
>
> If your EditPart is able to identify when this happens, it could force a
> refreshChildren on it's parent to add / remove itself:
>
> i.e.
> EditPart parent = getParent();
> parent removeChild(this);
> parent refreshChildren();
>
> -Steve
>
> "Christian Sell" <christian.sell@netcologne.de> wrote in message
> news:drsju9$g13$1@utils.eclipse.org...
>> Hello all,
>>
>> in our GEF applications the model objects override the equals() and
>> hashCode() methods to compute them based on certain object attributes.
>> GEF maintains an EditPart registry which uses these objects as keys,
>> which becomes a problem if the identity attributes of the model objects
>> change.
>>
>> My question: is there any recommended practice for cases like the one
>> above?
>>
>> thanks,
>> Christian
>
>
|
|
|
| Re: model objects changing their identity [message #208736 is a reply to message #208464] |
Wed, 08 February 2006 10:50 |
| Eclipse User |
|
|
|
Originally posted by: ingo.koch[nospam].sap.com
Hi Christian,
perhaps you should re-think your equals()-design.
Are two objects of your model, which are equal, really identical, one and
the same?
If not, you should not override the equals-method but test the "equality" of
your models in another method.
Regards, Ingo
"Christian Sell" <christian.sell@netcologne.de> wrote in message
news:dru17a$cbc$1@utils.eclipse.org...
> thats what I concluded, too. I guess the real lesson is that the model
> should not override the concept of equality/identity at all.
>
> christian
>
> Steven Shaw wrote:
> > I would expect that you would have to remove and then re-add the model
> > objects from the registry under these conditions.
> >
> > If your EditPart is able to identify when this happens, it could force a
> > refreshChildren on it's parent to add / remove itself:
> >
> > i.e.
> > EditPart parent = getParent();
> > parent removeChild(this);
> > parent refreshChildren();
> >
> > -Steve
> >
> > "Christian Sell" <christian.sell@netcologne.de> wrote in message
> > news:drsju9$g13$1@utils.eclipse.org...
> >> Hello all,
> >>
> >> in our GEF applications the model objects override the equals() and
> >> hashCode() methods to compute them based on certain object attributes.
> >> GEF maintains an EditPart registry which uses these objects as keys,
> >> which becomes a problem if the identity attributes of the model objects
> >> change.
> >>
> >> My question: is there any recommended practice for cases like the one
> >> above?
> >>
> >> thanks,
> >> Christian
> >
> >
|
|
|
| Re: model objects changing their identity [message #208768 is a reply to message #208736] |
Wed, 08 February 2006 13:09 |
| Eclipse User |
|
|
|
Originally posted by: christian.sell.netcologne.de
sure, I agree. Maybe overriding equals() and hashCode() is something one
should be more wary of doing in general. I went ahead and removed the
methods, and replaced them with an explicit comparison method wherever
needed - as you suggest.
One intention with this post was to make this requirement public.
-C
Ingo Koch wrote:
> Hi Christian,
>
> perhaps you should re-think your equals()-design.
> Are two objects of your model, which are equal, really identical, one and
> the same?
> If not, you should not override the equals-method but test the "equality" of
> your models in another method.
>
> Regards, Ingo
>
>
>
> "Christian Sell" <christian.sell@netcologne.de> wrote in message
> news:dru17a$cbc$1@utils.eclipse.org...
>> thats what I concluded, too. I guess the real lesson is that the model
>> should not override the concept of equality/identity at all.
>>
>> christian
>>
>> Steven Shaw wrote:
>>> I would expect that you would have to remove and then re-add the model
>>> objects from the registry under these conditions.
>>>
>>> If your EditPart is able to identify when this happens, it could force a
>>> refreshChildren on it's parent to add / remove itself:
>>>
>>> i.e.
>>> EditPart parent = getParent();
>>> parent removeChild(this);
>>> parent refreshChildren();
>>>
>>> -Steve
>>>
>>> "Christian Sell" <christian.sell@netcologne.de> wrote in message
>>> news:drsju9$g13$1@utils.eclipse.org...
>>>> Hello all,
>>>>
>>>> in our GEF applications the model objects override the equals() and
>>>> hashCode() methods to compute them based on certain object attributes.
>>>> GEF maintains an EditPart registry which uses these objects as keys,
>>>> which becomes a problem if the identity attributes of the model objects
>>>> change.
>>>>
>>>> My question: is there any recommended practice for cases like the one
>>>> above?
>>>>
>>>> thanks,
>>>> Christian
>>>
>
>
|
|
|
| Re: model objects changing their identity [message #208798 is a reply to message #208768] |
Wed, 08 February 2006 17:51 |
| Eclipse User |
|
|
|
Originally posted by: christian.sell.netcologne.de
in fact, I would generalize my rule like this:
overriding equals() and hashCode() based on mutable attributes (i.e.
ones that are not declared final) is *always* questionable
-christian
Christian Sell wrote:
> sure, I agree. Maybe overriding equals() and hashCode() is something one
> should be more wary of doing in general. I went ahead and removed the
> methods, and replaced them with an explicit comparison method wherever
> needed - as you suggest.
>
> One intention with this post was to make this requirement public.
>
> -C
>
> Ingo Koch wrote:
>> Hi Christian,
>>
>> perhaps you should re-think your equals()-design.
>> Are two objects of your model, which are equal, really identical, one and
>> the same?
>> If not, you should not override the equals-method but test the "equality" of
>> your models in another method.
>>
>> Regards, Ingo
>>
>>
>>
>> "Christian Sell" <christian.sell@netcologne.de> wrote in message
>> news:dru17a$cbc$1@utils.eclipse.org...
>>> thats what I concluded, too. I guess the real lesson is that the model
>>> should not override the concept of equality/identity at all.
>>>
>>> christian
>>>
>>> Steven Shaw wrote:
>>>> I would expect that you would have to remove and then re-add the model
>>>> objects from the registry under these conditions.
>>>>
>>>> If your EditPart is able to identify when this happens, it could force a
>>>> refreshChildren on it's parent to add / remove itself:
>>>>
>>>> i.e.
>>>> EditPart parent = getParent();
>>>> parent removeChild(this);
>>>> parent refreshChildren();
>>>>
>>>> -Steve
>>>>
>>>> "Christian Sell" <christian.sell@netcologne.de> wrote in message
>>>> news:drsju9$g13$1@utils.eclipse.org...
>>>>> Hello all,
>>>>>
>>>>> in our GEF applications the model objects override the equals() and
>>>>> hashCode() methods to compute them based on certain object attributes.
>>>>> GEF maintains an EditPart registry which uses these objects as keys,
>>>>> which becomes a problem if the identity attributes of the model objects
>>>>> change.
>>>>>
>>>>> My question: is there any recommended practice for cases like the one
>>>>> above?
>>>>>
>>>>> thanks,
>>>>> Christian
>>
|
|
|
Powered by
FUDForum. Page generated in 0.03644 seconds
] 
Search
Help
Register
Login
Home
