| final hashCode() and equals() in CDOObjectImpl, proxies and EMaps [message #1098887] |
Sat, 31 August 2013 17:17  |
Eclipse User |
|
|
|
Hello,
I have a model which uses CDOObject reference as an EMap key. For me it is a way to implement association class. When I put entries to that map I use proxies with unique identifiers, which are transparently resolved when different fragments of the model are put together.
Everything works perfectly fine except EMap.get(Object) method. The reason for this is that CDO EMap implementation used hashCode(). hashCode() is final in CDOObjectImpl and I can't override it and return hashCode of the ID attribute. Therefore, while CDOObject complies with the EObject contract - identity-based hashCode and equals, it becomes a predicament when it is used as a key in EMap.
My workaround for this was to create a filter class which implements EMap and implement get() and put() methods so they don't rely on hashCode(). I think this should be addressed in the framework itself - if EMap key is an instance of of EObject then the map should not rely on hash table in get() and put() methods.
Thoughts?
---
Best regards, Pavel Vlasov.
|
|
|
| Re: final hashCode() and equals() in CDOObjectImpl, proxies and EMaps [message #1099135 is a reply to message #1098887] |
Sun, 01 September 2013 02:55  |
| Eclipse User |
|
|
|
Hi Pavel,
I don't fully understand what you're suggesting. It would be best that you submit a bugzilla and attach a test case that
demos the wrong behaviour and your patch that is supposed to fix it. Please copy one of the existing tests as a start
point and try to use our test models.
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
Am 31.08.2013 23:17, schrieb Pavel Vlasov:
> Hello,
>
> I have a model which uses CDOObject reference as an EMap key. For me it is a way to implement association class. When
> I put entries to that map I use proxies with unique identifiers, which are transparently resolved when different
> fragments of the model are put together.
> Everything works perfectly fine except EMap.get(Object) method. The reason for this is that CDO EMap implementation
> used hashCode(). hashCode() is final in CDOObjectImpl and I can't override it and return hashCode of the ID attribute.
> Therefore, while CDOObject complies with the EObject contract - identity-based hashCode and equals, it becomes a
> predicament when it is used as a key in EMap.
>
> My workaround for this was to create a filter class which implements EMap and implement get() and put() methods so
> they don't rely on hashCode(). I think this should be addressed in the framework itself - if EMap key is an instance
> of of EObject then the map should not rely on hash table in get() and put() methods.
>
> Thoughts?
> ---
> Best regards, Pavel Vlasov.
|
|
|
Powered by
FUDForum. Page generated in 0.04273 seconds
Home