Home » Modeling » EMF » [CDO] Problems with eAnnotation
|
| Re: [CDO] Problems with eAnnotation [message #645309 is a reply to message #645304] |
Thu, 16 December 2010 14:58   |
|
This is a multi-part message in MIME format.
--------------030205080302010901000909
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Hi Stephane,
It seems to me you don't have a problem caused by annotations but rather a problem with modelevolution, which is generally not yet supported by CDO. Adding anything, including annotations, to a model results in a different model, an evolved model. I expect to come up with a solution to this complex problem in the first quarter of 2011.
In the meatime you'de have to migrate your existing data manually, possibly with the brand new export/import feature:
330235: Backup/Restore Functionality
https://bugs.eclipse.org/bugs/show_bug.cgi?id=330235
332355: Provide OSGi console commands to start/stop/list/export/import repositories
https://bugs.eclipse.org/bugs/show_bug.cgi?id=332355
332428: CDO Export does not support custom datatypes
https://bugs.eclipse.org/bugs/show_bug.cgi?id=332428
Just the other second I answered some related questions of a customer. Please find an excerpt attached.
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
Am 16.12.2010 15:40, schrieb Stephane Rousseau:
> Hello every one,
>
> As said in the subject, I have problems managing eAnnotation in my ecore metamodel with CDO.
>
> I have an ecore metamodel. In this metamodel, I have one eAnnotation per eClassifier for documentation purpose :
>
> <eClassifiers xsi:type="ecore:EClass" name="MyFirstClass" eSuperTypes="#//ModelElement">
> <eAnnotations source="http://www.eclipse.org/emf/2002/GenModel">
> <details key="documentation" value="a Class"/>
> </eAnnotations>
> <eStructuralFeatures xsi:type="ecore:EAttribute" name="myFirstAttribute" eType="ecore:EDataType http://www.eclipse.org/emf/2002/Ecore#//EString"/>
>
> I used the tutoriual here ( http://wiki.eclipse.org/FAQ_for_CDO_and_Net4j#What.27s_the_b est_way_to_convert_an_XMIResource_to_a_CDOResource.3F) to have my ecore + CDO metamodel
>
> After reading my xmi file, at the commit time, I have this exception in mapMetaInstances(EPackage ePackage, CDOIDMetaRange metaIDRange) method :
>
> throw new IllegalStateException("range.size() != metaIDRange.size()");
>
> The difference is the sum of eAnnotation occurrences and eStringToStrigMapEntity occurrences.
>
> If I remove the eAnnnotation from my ecore metamodel, I have no more the error.
>
> So, I wonder if I make something wrong (not impossible at all) or is there any bug around this ?
>
> Here is my environment : Eclipse SDK 3.7.0 + CDO 4.0.0.v20101108-0809
>
> Regards
>
> Stephane Rousseau
--------------030205080302010901000909
Content-Type: text/plain;
name="mail.txt"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
filename="mail.txt"
PiBRdWVzdGlvbnMgSSBjdXJyZW50bHkgaGF2ZSBhcmU6DQo+IC0gSG93IHdl IGNhbiByZWFk
IHRoZSBleHBvcnRlZCBmaWxlPw0KRWl0aGVyIHdpdGggdGhlIHN1cHBsZW1l bnRhbCBDRE8g
aW1wb3J0ZXIgb3Igd2l0aCBhbnkgb3RoZXIgWE1MIHBhcnNlci90cmFuc2Zv cm1lci4NCg0K
PiAgICBJcyB0aGlzIGRvbmUgdGhyb3VnaCBFTUYsDQpOby4gV2VsbCwgaXQg Y291bGQgYmUs
IGJlY2F1c2UgdmlydHVhbGx5IGV2ZXJ5IFhNTCBTY2hlbWEgaXMgY29udmVy dGlibGUgdG8g
YW4gRWNvcmUgbW9kZWwuIEJ1dCB0aGF0IHdvdWxkbid0IHNjYWxlIHdlbGwu IFRoYXQgd2Fz
IHRoZSByZWFzb24gZm9yIG15IGRlY2lzaW9uIG5vdCB0byB1c2UgYW4gRU1G IHJlc291cmNl
IGluIHRoZSBmaXJzdCBwbGFjZS4NCg0KPiAgIEkgaXQgYW4gRU1GIHJlc291 cmNlPw0KTm8u
IFNlZSBhYm92ZS4NCg0KPiBPciBpcyBpdCByZWFkIHVzaW5nIHN0cmFpZ2h0 IFhNTD8NClll
cy4NCg0KPiAtIEhvdyB3ZSBjYW4gd3JpdGUgYSBmaWxlIHdoaWNoIGlzIGNv cnJlY3QgdG8g
aW1wb3J0LiBEbyB3ZSBuZWVkIHRvIHdyaXRlIGlzIHVzaW5nIEVNRj8gIE9y IHVzaW5nIHN0
cmFpZ2h0IFhNTD8NClRleHQgKFJlZ0V4Pykgb3IgWE1MLiBTZWUgYWJvdmUu DQoNCj4gLSBJ
ZiB3ZSB3cml0ZSB0aGUgZmlsZSBvdXJzZWx2ZXMgZHVyaW5nIGEgbWlncmF0 aW9uLCB3aGF0
IHZhbGlkYXRpb24gaXMgbmVlZGVkIG9uIHRoZSBmaWxlIHRvIGVuc3VyZSBD RG8gY2FuIHJl
YWQgaXQgYWdhaW4/DQpHb29kIHF1ZXN0aW9uLiBTaW1wbGUgYW5zd2VyOiBU aGUgQ0RPIGlt
cG9ydGVyIG11c3QgYmUgYWJsZSB0byByZWFkIGl0IHdpdGhvdXQgZXJyb3Jz LiBTbywgdGhh
dCdzIG9uZSBraW5kIG9mIHZhbGlkYXRpb24uIEkgY291bGQgdGhpbmsgYWJv dXQgYW4gaW1w
b3J0ZXIgb3B0aW9uIGxpa2UgIi12YWxpZGF0ZU9ubHk9dHJ1ZSIgb3Igc28u IEJ1dCBJIGRv
dWJ0IHRoYXQgdGhpcyBpcyBhIGZldyBob3VycyBqb2IuDQoNCj4gLSBXaGF0 IGlzIHRoZSBi
ZXN0IHBsYWNlLA0KSSBkb24ndCB1bmRlcnN0YW5kIHRoaXMgcXVlc3Rpb24u DQoNCj4gdGhl
IGJlc3Qgd2F5IHRvIGFjdHVhbGx5IHBlcmZvcm0gdGhlIG1pZ3JhdGlvbiBs b2dpYy4NCkhh
cmQgcXVlc3Rpb24uIEkgZ3Vlc3MgaXQgbWFqb3JseSBkZXBlbmRzIG9uIHRo ZSBhdmFpbGFi
bGUga25vd2xlZGdlLiBlLmcuIGRvIHlvdSBoYXZlIFhTTFQgZXhwZXJ0aXNl LCBldGMuDQoN
Ck9uZSBtb3JlIG5vdGU6ICBBIENETyB0ZWFtIG1lbWJlciBpcyBub3cgdHJ5 aW5nIHRvIGVt
cGxveSB0aGUgbmV3IGltcG9ydC9leHBvcnQgZmVhdHVyZSBmb3IgdGhlIGFs cmVhZHkgbWVu
dGlvbmVkIGZyYW1ld29yayBldm9sdXRpb24gc3VwcG9ydC4gVGhlcmUncyBv bmUgcGllY2Ug
b2YgaW5mb3JtYXRpb24gaW4gdGhlIGV4cG9ydCBmaWxlIHRoYXQgdHVybmVk IG91dCB0byBi
ZSBwcm9ibGVtYXRpYy4gRWFjaCByZWdpc3RlcmVkIE1vZGVsIChpbmNsLiB0 aGUgc3lzdGVt
IG1vZGVscyBFY29yZSwgRXJlc291cmNlIGFuZCBFdHlwZXMpIGlzIGFzc29j aWF0ZWQgd2l0
aCBhICJyYW5nZSBvZiBNZXRhIElEcyIuIElEcyBmcm9tIHRoZXNlIHJhbmdl cyBhcmUgYXNz
b2NpYXRlZCB3aXRoIHRoZSBFTW9kZWxFbGVtZW50cyBjb250YWluZWQgYnkg dGhlc2UgbW9k
ZWxzIGZvciBzZXZlcmFsIHB1cnBvc2VzOg0KDQphKSBUaGV5IGVuYWJsZSAi bm9ybWFsIiBp
bnN0YW5jZXMgdG8gcG9pbnQgdG8gbW9kZWwgZWxlbWVudHMgdGhyb3VnaCBD RE9JRHMgKE1l
dGEgSURzKQ0KYikgVGhleSBlbmFibGUgdGhlIGZyYW1ld29yayB0byBkZXRl Y3QgaWxsZWdh
bCBtb2RlbCBldm9sdXRpb25zDQpjKSBUaGV5IGFyZSB1c2VkIGJ5IGNlcnRh aW4gSVN0b3Jl
cyBmb3IgaW50ZXJuYWwgcHVycG9zZXMgKG9wdGltaXphdGlvbiwgcmVtb3Zp bmcgY29sdW1u
IG5hbWUgYW1iaWd1aXRpZXMsIGV0Yy4uLikNCg0KRm9yIHlvdSB0aGF0IG1l YW5zIG9uZSBv
ciB0d28gY2hhbGxlbmdlczoNCg0KSWYgeW91ciAidHJhbnNmb3JtYXRpb24i IG9mIGEgbWV0
YSBtb2RlbCBjaGFuZ2VzIHRoZSBzaXplIG9mIGFueSBtZXRhIG1vZGVsIChu dW1iZXIgb2Yg
RU1vZGVsRWxlbWVudHMpIHlvdSdsbCBuZWVkIHRvIGVuc3VyZSB0aGF0IHBy b3BlciAoY29u
c2VjdXRpdmUsIG5vbi1vdmVybGFwcGluZykgTWV0YSBJRCByYW5nZXMgYXJl IGFwcGxpZWQg
dG8gdGhlIHJlc3VsdGluZyBYTUwuDQoNCk9wdGlvbmFsbHksIGlmIHlvdSdy ZSB1c2luZyBz
b21lIE1ldGEgSURzIGluIHlvdXIgb3duIGluc3RhbmNlcyAoZS5nLiBoYXZl IGEgcmVmZXJl
bmNlIGZyb20gYSBNYXJrZXRhYmxlUHJvZHVjdCB0byBhbiBFQ2xhc3MpIHRo ZW4geW91IG5l
ZWQgdG8gbWlncmF0ZSB0aGVzZSB0YXJnZXQgSURzIGFjY29yZGluZyB0byB0 aGUgbmV3IHJh
bmdlcy4NCg0KSSdtIGN1cnJlbnRseSBpbiBhIGRpc2N1c3Npb24gd2l0aCBt eSB0ZWFtIHdo
ZXRoZXIgd2Ugc2hvdWxkIGNvbnNpZGVyIHRoZXNlIE1ldGEgSUQgdGhpbmdz IGFuIGltcGxl
bWVudGF0aW9uIGRldGFpbCBpbiB0aGUgc2NvcGUgb2YgdGhlIGltcG9ydC9l eHBvcnQgbWVj
aGFuc2ltLiBUaGVyZSdzIGEgY2hhbmNlIHRoYXQgd2UgY2FuIHJlbW92ZSB0 aGlzIHR5cGUg
b2YgaW5mb3JtYXRpb24gZnJvbXRoZSBleHBvcnQgZmlsZSBhbmQgcmVwbGFj ZSB5b3VyIG93
biBNZXRhIElEIHJlZmVyZW5jZXMgYnkgbm9ybWFsIEVNRiBwcm94eSBVUklz IGludG8gdGhl
IHJlc3BlY3RpdmUgbWV0YSBtb2RlbHMuIFRoaXMgZGlzY3Vzc2lvbiBpcyBu b3QgZmluaXNo
ZWQsIHlldC4NCg0KU29ycnksIEkga25vdyB0aGlzIGlzIGEgbG90IG9mIHN0 dWZmIHRvIGNv
bnNpZGVyIGZvciB5b3UsIGJ1dCBJJ3ZlIG5ldmVyIHN1Z2dlc3RlZCB0aGF0 IHRoaXMgZXhw
b3J0L2ltcG9ydCBtZWNoYW5zaW0gaXN0IHRoZSAqYmVzdCogc29sdXRpb24g dG8gbW9kZWwg
ZXZvbHV0aW9uLiA=
--------------030205080302010901000909--
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
|
|
|
| Re: [CDO] Problems with eAnnotation [message #645313 is a reply to message #645309] |
Thu, 16 December 2010 15:08 |
 Stephane Rousseau
Messages: 34
Registered: September 2010 |
Member |
|
|
Thank you for your quick answer, Eike,
Before making the post, I read some posts about model evolution. But I
think it's not my case.
The xmi file is conformed to the ecore metamodel with annotation even if
there is no annotation in my xmi file.
The only evolution I've made is to remove eAnnotation from the ecore
metamodel to complete the commit.
I will read your links to see if the situations are the same, and for my
personal information too.
Regards
Stephane
On 16/12/2010 15:58, Eike Stepper wrote:
> Hi Stephane,
>
> It seems to me you don't have a problem caused by annotations but rather
> a problem with modelevolution, which is generally not yet supported by
> CDO. Adding anything, including annotations, to a model results in a
> different model, an evolved model. I expect to come up with a solution
> to this complex problem in the first quarter of 2011.
>
> In the meatime you'de have to migrate your existing data manually,
> possibly with the brand new export/import feature:
>
> 330235: Backup/Restore Functionality
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=330235
>
> 332355: Provide OSGi console commands to start/stop/list/export/import
> repositories
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332355
>
> 332428: CDO Export does not support custom datatypes
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332428
>
> Just the other second I answered some related questions of a customer.
> Please find an excerpt attached.
>
> Cheers
> /Eike
>
> ----
> http://www.esc-net.de
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>
> Am 16.12.2010 15:40, schrieb Stephane Rousseau:
>> Hello every one,
>>
>> As said in the subject, I have problems managing eAnnotation in my
>> ecore metamodel with CDO.
>>
>> I have an ecore metamodel. In this metamodel, I have one eAnnotation
>> per eClassifier for documentation purpose :
>>
>> <eClassifiers xsi:type="ecore:EClass" name="MyFirstClass"
>> eSuperTypes="#//ModelElement">
>> <eAnnotations source="http://www.eclipse.org/emf/2002/GenModel">
>> <details key="documentation" value="a Class"/>
>> </eAnnotations>
>> <eStructuralFeatures xsi:type="ecore:EAttribute"
>> name="myFirstAttribute" eType="ecore:EDataType
>> http://www.eclipse.org/emf/2002/Ecore#//EString"/>
>>
>> I used the tutoriual here
>> ( http://wiki.eclipse.org/FAQ_for_CDO_and_Net4j#What.27s_the_b est_way_to_convert_an_XMIResource_to_a_CDOResource.3F)
>> to have my ecore + CDO metamodel
>>
>> After reading my xmi file, at the commit time, I have this exception
>> in mapMetaInstances(EPackage ePackage, CDOIDMetaRange metaIDRange)
>> method :
>>
>> throw new IllegalStateException("range.size() != metaIDRange.size()");
>>
>> The difference is the sum of eAnnotation occurrences and
>> eStringToStrigMapEntity occurrences.
>>
>> If I remove the eAnnnotation from my ecore metamodel, I have no more
>> the error.
>>
>> So, I wonder if I make something wrong (not impossible at all) or is
>> there any bug around this ?
>>
>> Here is my environment : Eclipse SDK 3.7.0 + CDO 4.0.0.v20101108-0809
>>
>> Regards
>>
>> Stephane Rousseau
|
|
|
| Re: [CDO] Problems with eAnnotation [message #645314 is a reply to message #645309] |
Thu, 16 December 2010 15:15 |
Stephane Rousseau
Messages: 34
Registered: September 2010 |
Member |
|
|
Just one more thing,
I commit into a new database after each reading so the database is
always consistent with the metamodel.
Stephane
On 16/12/2010 15:58, Eike Stepper wrote:
> Hi Stephane,
>
> It seems to me you don't have a problem caused by annotations but rather
> a problem with modelevolution, which is generally not yet supported by
> CDO. Adding anything, including annotations, to a model results in a
> different model, an evolved model. I expect to come up with a solution
> to this complex problem in the first quarter of 2011.
>
> In the meatime you'de have to migrate your existing data manually,
> possibly with the brand new export/import feature:
>
> 330235: Backup/Restore Functionality
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=330235
>
> 332355: Provide OSGi console commands to start/stop/list/export/import
> repositories
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332355
>
> 332428: CDO Export does not support custom datatypes
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332428
>
> Just the other second I answered some related questions of a customer.
> Please find an excerpt attached.
>
> Cheers
> /Eike
>
> ----
> http://www.esc-net.de
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>
> Am 16.12.2010 15:40, schrieb Stephane Rousseau:
>> Hello every one,
>>
>> As said in the subject, I have problems managing eAnnotation in my
>> ecore metamodel with CDO.
>>
>> I have an ecore metamodel. In this metamodel, I have one eAnnotation
>> per eClassifier for documentation purpose :
>>
>> <eClassifiers xsi:type="ecore:EClass" name="MyFirstClass"
>> eSuperTypes="#//ModelElement">
>> <eAnnotations source="http://www.eclipse.org/emf/2002/GenModel">
>> <details key="documentation" value="a Class"/>
>> </eAnnotations>
>> <eStructuralFeatures xsi:type="ecore:EAttribute"
>> name="myFirstAttribute" eType="ecore:EDataType
>> http://www.eclipse.org/emf/2002/Ecore#//EString"/>
>>
>> I used the tutoriual here
>> ( http://wiki.eclipse.org/FAQ_for_CDO_and_Net4j#What.27s_the_b est_way_to_convert_an_XMIResource_to_a_CDOResource.3F)
>> to have my ecore + CDO metamodel
>>
>> After reading my xmi file, at the commit time, I have this exception
>> in mapMetaInstances(EPackage ePackage, CDOIDMetaRange metaIDRange)
>> method :
>>
>> throw new IllegalStateException("range.size() != metaIDRange.size()");
>>
>> The difference is the sum of eAnnotation occurrences and
>> eStringToStrigMapEntity occurrences.
>>
>> If I remove the eAnnnotation from my ecore metamodel, I have no more
>> the error.
>>
>> So, I wonder if I make something wrong (not impossible at all) or is
>> there any bug around this ?
>>
>> Here is my environment : Eclipse SDK 3.7.0 + CDO 4.0.0.v20101108-0809
>>
>> Regards
>>
>> Stephane Rousseau
|
|
|
| Re: [CDO] Problems with eAnnotation [message #645332 is a reply to message #645313] |
Thu, 16 December 2010 15:40 |
|
Am 16.12.2010 16:08, schrieb Stephane Rousseau:
> Thank you for your quick answer, Eike,
>
> Before making the post, I read some posts about model evolution. But I think it's not my case.
And I think it is :P
> The xmi file is conformed to the ecore metamodel with annotation even if there is no annotation in my xmi file.
A particular persistence mechanism is always free to consider particular model evolutions as compatible or not, or even try to implicitely migrate instances during load or save. The following assertion is false:
(instance conforms to model1 and instance conforms to model2) implies model1 == model2
>
> The only evolution I've made is to remove eAnnotation from the ecore metamodel
There we go!
> to complete the commit.
>
> I will read your links to see if the situations are the same, and for my personal information too.
Please also understand that for certain persistence formats like XML or XMI it is just easier to cope with compatible evolution. While other formats like SQL tables are way harder to adjust. In the case of CDO there's even not a single implied format because custom data backends can be plugged into the framework.
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
>
> Regards
>
> Stephane
>
>
> On 16/12/2010 15:58, Eike Stepper wrote:
>> Hi Stephane,
>>
>> It seems to me you don't have a problem caused by annotations but rather
>> a problem with modelevolution, which is generally not yet supported by
>> CDO. Adding anything, including annotations, to a model results in a
>> different model, an evolved model. I expect to come up with a solution
>> to this complex problem in the first quarter of 2011.
>>
>> In the meatime you'de have to migrate your existing data manually,
>> possibly with the brand new export/import feature:
>>
>> 330235: Backup/Restore Functionality
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=330235
>>
>> 332355: Provide OSGi console commands to start/stop/list/export/import
>> repositories
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332355
>>
>> 332428: CDO Export does not support custom datatypes
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332428
>>
>> Just the other second I answered some related questions of a customer.
>> Please find an excerpt attached.
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://www.esc-net.de
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>>
>> Am 16.12.2010 15:40, schrieb Stephane Rousseau:
>>> Hello every one,
>>>
>>> As said in the subject, I have problems managing eAnnotation in my
>>> ecore metamodel with CDO.
>>>
>>> I have an ecore metamodel. In this metamodel, I have one eAnnotation
>>> per eClassifier for documentation purpose :
>>>
>>> <eClassifiers xsi:type="ecore:EClass" name="MyFirstClass"
>>> eSuperTypes="#//ModelElement">
>>> <eAnnotations source="http://www.eclipse.org/emf/2002/GenModel">
>>> <details key="documentation" value="a Class"/>
>>> </eAnnotations>
>>> <eStructuralFeatures xsi:type="ecore:EAttribute"
>>> name="myFirstAttribute" eType="ecore:EDataType
>>> http://www.eclipse.org/emf/2002/Ecore#//EString"/>
>>>
>>> I used the tutoriual here
>>> ( http://wiki.eclipse.org/FAQ_for_CDO_and_Net4j#What.27s_the_b est_way_to_convert_an_XMIResource_to_a_CDOResource.3F)
>>> to have my ecore + CDO metamodel
>>>
>>> After reading my xmi file, at the commit time, I have this exception
>>> in mapMetaInstances(EPackage ePackage, CDOIDMetaRange metaIDRange)
>>> method :
>>>
>>> throw new IllegalStateException("range.size() != metaIDRange.size()");
>>>
>>> The difference is the sum of eAnnotation occurrences and
>>> eStringToStrigMapEntity occurrences.
>>>
>>> If I remove the eAnnnotation from my ecore metamodel, I have no more
>>> the error.
>>>
>>> So, I wonder if I make something wrong (not impossible at all) or is
>>> there any bug around this ?
>>>
>>> Here is my environment : Eclipse SDK 3.7.0 + CDO 4.0.0.v20101108-0809
>>>
>>> Regards
>>>
>>> Stephane Rousseau
>
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
|
|
|
| Re: [CDO] Problems with eAnnotation [message #645333 is a reply to message #645314] |
Thu, 16 December 2010 15:46 |
|
Am 16.12.2010 16:15, schrieb Stephane Rousseau:
> Just one more thing,
>
> I commit into a new database after each reading so the database is always consistent with the metamodel.
Well, that indeed doesn't sound like a model evolution problem then, because from the point of the new repository it's just the first and only version.
It rather sounds like a deployment problem on your server. Please note that the package registry of a CDO repository by default consults the global EMF package registry before it attempts to create a dynamic package from the stored information about the originally committed package. In other words, generated packages have precedence.
It seems that you have old package versions registered in your global EMF package registry, Older and different from the ones deployed to and used by your clients.
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
>
> Stephane
>
> On 16/12/2010 15:58, Eike Stepper wrote:
>> Hi Stephane,
>>
>> It seems to me you don't have a problem caused by annotations but rather
>> a problem with modelevolution, which is generally not yet supported by
>> CDO. Adding anything, including annotations, to a model results in a
>> different model, an evolved model. I expect to come up with a solution
>> to this complex problem in the first quarter of 2011.
>>
>> In the meatime you'de have to migrate your existing data manually,
>> possibly with the brand new export/import feature:
>>
>> 330235: Backup/Restore Functionality
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=330235
>>
>> 332355: Provide OSGi console commands to start/stop/list/export/import
>> repositories
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332355
>>
>> 332428: CDO Export does not support custom datatypes
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332428
>>
>> Just the other second I answered some related questions of a customer.
>> Please find an excerpt attached.
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://www.esc-net.de
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>>
>> Am 16.12.2010 15:40, schrieb Stephane Rousseau:
>>> Hello every one,
>>>
>>> As said in the subject, I have problems managing eAnnotation in my
>>> ecore metamodel with CDO.
>>>
>>> I have an ecore metamodel. In this metamodel, I have one eAnnotation
>>> per eClassifier for documentation purpose :
>>>
>>> <eClassifiers xsi:type="ecore:EClass" name="MyFirstClass"
>>> eSuperTypes="#//ModelElement">
>>> <eAnnotations source="http://www.eclipse.org/emf/2002/GenModel">
>>> <details key="documentation" value="a Class"/>
>>> </eAnnotations>
>>> <eStructuralFeatures xsi:type="ecore:EAttribute"
>>> name="myFirstAttribute" eType="ecore:EDataType
>>> http://www.eclipse.org/emf/2002/Ecore#//EString"/>
>>>
>>> I used the tutoriual here
>>> ( http://wiki.eclipse.org/FAQ_for_CDO_and_Net4j#What.27s_the_b est_way_to_convert_an_XMIResource_to_a_CDOResource.3F)
>>> to have my ecore + CDO metamodel
>>>
>>> After reading my xmi file, at the commit time, I have this exception
>>> in mapMetaInstances(EPackage ePackage, CDOIDMetaRange metaIDRange)
>>> method :
>>>
>>> throw new IllegalStateException("range.size() != metaIDRange.size()");
>>>
>>> The difference is the sum of eAnnotation occurrences and
>>> eStringToStrigMapEntity occurrences.
>>>
>>> If I remove the eAnnnotation from my ecore metamodel, I have no more
>>> the error.
>>>
>>> So, I wonder if I make something wrong (not impossible at all) or is
>>> there any bug around this ?
>>>
>>> Here is my environment : Eclipse SDK 3.7.0 + CDO 4.0.0.v20101108-0809
>>>
>>> Regards
>>>
>>> Stephane Rousseau
>
Cheers
/Eike
----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper
|
|
|
| Re: [CDO] Problems with eAnnotation [message #645493 is a reply to message #645333] |
Fri, 17 December 2010 13:25 |
Stephane Rousseau
Messages: 34
Registered: September 2010 |
Member |
|
|
Hello,
As I am in the prototyping part, I will not spend much time to see where
the problem comes from.
So I will be back with my problem in some weeks (in 2011).
Happy new year.
Stephane
On 16/12/2010 16:46, Eike Stepper wrote:
> Am 16.12.2010 16:15, schrieb Stephane Rousseau:
>> Just one more thing,
>>
>> I commit into a new database after each reading so the database is
>> always consistent with the metamodel.
> Well, that indeed doesn't sound like a model evolution problem then,
> because from the point of the new repository it's just the first and
> only version.
>
> It rather sounds like a deployment problem on your server. Please note
> that the package registry of a CDO repository by default consults the
> global EMF package registry before it attempts to create a dynamic
> package from the stored information about the originally committed
> package. In other words, generated packages have precedence.
>
> It seems that you have old package versions registered in your global
> EMF package registry, Older and different from the ones deployed to and
> used by your clients.
>
> Cheers
> /Eike
>
> ----
> http://www.esc-net.de
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>>
>> Stephane
>>
>> On 16/12/2010 15:58, Eike Stepper wrote:
>>> Hi Stephane,
>>>
>>> It seems to me you don't have a problem caused by annotations but rather
>>> a problem with modelevolution, which is generally not yet supported by
>>> CDO. Adding anything, including annotations, to a model results in a
>>> different model, an evolved model. I expect to come up with a solution
>>> to this complex problem in the first quarter of 2011.
>>>
>>> In the meatime you'de have to migrate your existing data manually,
>>> possibly with the brand new export/import feature:
>>>
>>> 330235: Backup/Restore Functionality
>>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=330235
>>>
>>> 332355: Provide OSGi console commands to start/stop/list/export/import
>>> repositories
>>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332355
>>>
>>> 332428: CDO Export does not support custom datatypes
>>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=332428
>>>
>>> Just the other second I answered some related questions of a customer.
>>> Please find an excerpt attached.
>>>
>>> Cheers
>>> /Eike
>>>
>>> ----
>>> http://www.esc-net.de
>>> http://thegordian.blogspot.com
>>> http://twitter.com/eikestepper
>>>
>>>
>>>
>>> Am 16.12.2010 15:40, schrieb Stephane Rousseau:
>>>> Hello every one,
>>>>
>>>> As said in the subject, I have problems managing eAnnotation in my
>>>> ecore metamodel with CDO.
>>>>
>>>> I have an ecore metamodel. In this metamodel, I have one eAnnotation
>>>> per eClassifier for documentation purpose :
>>>>
>>>> <eClassifiers xsi:type="ecore:EClass" name="MyFirstClass"
>>>> eSuperTypes="#//ModelElement">
>>>> <eAnnotations source="http://www.eclipse.org/emf/2002/GenModel">
>>>> <details key="documentation" value="a Class"/>
>>>> </eAnnotations>
>>>> <eStructuralFeatures xsi:type="ecore:EAttribute"
>>>> name="myFirstAttribute" eType="ecore:EDataType
>>>> http://www.eclipse.org/emf/2002/Ecore#//EString"/>
>>>>
>>>> I used the tutoriual here
>>>> ( http://wiki.eclipse.org/FAQ_for_CDO_and_Net4j#What.27s_the_b est_way_to_convert_an_XMIResource_to_a_CDOResource.3F)
>>>>
>>>> to have my ecore + CDO metamodel
>>>>
>>>> After reading my xmi file, at the commit time, I have this exception
>>>> in mapMetaInstances(EPackage ePackage, CDOIDMetaRange metaIDRange)
>>>> method :
>>>>
>>>> throw new IllegalStateException("range.size() != metaIDRange.size()");
>>>>
>>>> The difference is the sum of eAnnotation occurrences and
>>>> eStringToStrigMapEntity occurrences.
>>>>
>>>> If I remove the eAnnnotation from my ecore metamodel, I have no more
>>>> the error.
>>>>
>>>> So, I wonder if I make something wrong (not impossible at all) or is
>>>> there any bug around this ?
>>>>
>>>> Here is my environment : Eclipse SDK 3.7.0 + CDO 4.0.0.v20101108-0809
>>>>
>>>> Regards
>>>>
>>>> Stephane Rousseau
>>
|
|
|
Goto Forum:
Current Time: Fri Apr 19 15:10:40 GMT 2024
Powered by FUDForum. Page generated in 0.02368 seconds |
] 
Search
Help
Register
Login
Home
