|
| Re: Loading unique reference [message #430942 is a reply to message #430939] |
Thu, 18 June 2009 22:47   |
 Ed Merks
Messages: 33133
Registered: July 2009 |
Senior Member |
|
|
Martynas,
Comments below.
Martynas Lelevicius wrote:
> Hi,
>
> On reference loading (XMI Resource) the uniqueness of EList elements
> is not preserved.
>
> Suppose EObject 'A' has a reference 'b' of type 'B' ('A' and 'B' are
> from different packages).
> EReference 'b' is unique and has a multiplicity of 0..* (isMany=true).
>
> Create A and B instances, reference them through 'b' and save to
> different (XMI)resources.
> Then load the resources (make A already reference B before loading,
> e.g. in A constructor add B to 'b').
EMF assumes instances are empty when first created, so this sounds bogus
> After the load 'A' references two the same 'B', despite the reference
> is unique.
The deserializer bypasses the duplicate checking for efficiency, so if
you do something bogus in the constructor, all bets are off.
>
> EMF 2.4.2.v200902171115
>
> Maybe the explanation is not very clear, so I am attaching the sample
> source.
>
> Thanks,
> Martynas
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
|
|
| Re: Loading unique reference [message #430948 is a reply to message #430947] |
Fri, 19 June 2009 08:46 |
Ed Merks
Messages: 33133
Registered: July 2009 |
Senior Member |
|
|
Martynas,
My impression is that it's not a default if it's runtime dependent, but
yes, additional factory methods that do specialized initialization makes
the most sense to me.
Martynas Lelevicius wrote:
> Hi,
>
> So what can be an alternative option for having runtime dependent
> "default" value:
> 1) Set value "manually" after object instantiation (e.g. create
> additional factory method)
> 2) ...?
>
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
|
Powered by
FUDForum. Page generated in 1.69683 seconds
] 
Search
Help
Register
Login
Home
