| Change of policy for UUIDs [message #535308] |
Sat, 22 May 2010 01:57  |
Eclipse User |
|
|
|
Originally posted by: atoulme.intalio.com
It looks like 2.6 is changing the way it handles UUIDs. It used to
create them by default and it's not the case anymore for the BPMN
modeler. Do I need to register a subclass of XMIResourceImpl somewhere
to enforce using them ? I am a bit clueless as to where the change
occurred, it looks rather subtle to me.
Is there a checkbox somewhere in genmodel that I missed ? Is there a doc
in the wiki for this change ? (EMF XMI is returning quite a few things).
Thanks,
Antoine
|
|
|
|
| Re: Change of policy for UUIDs [message #535343 is a reply to message #535334] |
Sat, 22 May 2010 14:31  |
| Eclipse User |
|
|
|
Originally posted by: atoulme.intalio.com
OK, I'll see what it takes and I will update the wiki if it makes any
sense for others. Thanks!
On 5/22/10 5:54 AM, Ed Merks wrote:
> Antoine,
>
> Comments below.
>
> Antoine Toulme wrote:
>> It looks like 2.6 is changing the way it handles UUIDs.
> I don't recall making any changes in UUID handling.
>> It used to create them by default and it's not the case anymore for
>> the BPMN modeler.
> The resource implementation is responsible for this.
>> Do I need to register a subclass of XMIResourceImpl somewhere to
>> enforce using them ?
> Yes.
>> I am a bit clueless as to where the change occurred, it looks rather
>> subtle to me.
> A specialized XMIResourceImpl that overrides useUUIDs to return true is
> needed.
>>
>> Is there a checkbox somewhere in genmodel that I missed ?
> No.
>> Is there a doc in the wiki for this change ? (EMF XMI is returning
>> quite a few things).
> No. You could add something to the FAQ or an EMF Recipe; both are in the
> wiki.
>>
>> Thanks,
>>
>> Antoine
|
|
|
| Re: Change of policy for UUIDs [message #535370 is a reply to message #535343] |
Sat, 22 May 2010 20:42 |
| Eclipse User |
|
|
|
Originally posted by: atoulme.intalio.com
My extension point to change the content type was wrong. It was not
loaded, hence the default resource factory was used, and I got myself in
trouble.
Even with that, I still had trouble creating the initial model. Turns
out, since my describer attached itself to only load documents with the
right namespace on the root element, it would not be selected when I
create the initial model which was empty. Workaround: make sure to
create the load myself instead of using the resourceSet.createResource
method.
Over all, I goofed, but I'm back on tracks. Sorry for pointing fingers
the wrong way...
On 5/22/10 7:31 AM, Antoine Toulme wrote:
> OK, I'll see what it takes and I will update the wiki if it makes any
> sense for others. Thanks!
> On 5/22/10 5:54 AM, Ed Merks wrote:
>> Antoine,
>>
>> Comments below.
>>
>> Antoine Toulme wrote:
>>> It looks like 2.6 is changing the way it handles UUIDs.
>> I don't recall making any changes in UUID handling.
>>> It used to create them by default and it's not the case anymore for
>>> the BPMN modeler.
>> The resource implementation is responsible for this.
>>> Do I need to register a subclass of XMIResourceImpl somewhere to
>>> enforce using them ?
>> Yes.
>>> I am a bit clueless as to where the change occurred, it looks rather
>>> subtle to me.
>> A specialized XMIResourceImpl that overrides useUUIDs to return true is
>> needed.
>>>
>>> Is there a checkbox somewhere in genmodel that I missed ?
>> No.
>>> Is there a doc in the wiki for this change ? (EMF XMI is returning
>>> quite a few things).
>> No. You could add something to the FAQ or an EMF Recipe; both are in the
>> wiki.
>>>
>>> Thanks,
>>>
>>> Antoine
>
|
|
|
Powered by
FUDForum. Page generated in 0.02292 seconds
] 
Search
Help
Register
Login
Home
