| [EMFT Compare] Additionnal ressource and compare [message #133076] |
Thu, 12 February 2009 15:16  |
Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
Hi all,
I have a UML model with a loaded and applied profile. I add a Class
property to this model and then I compare to the latest version. EMF
Compare find the difference but it also find as much difference as there
are stereotypes.
I think that is because it does not load the profile in the left model
before compare.
I tried to load another additional resource (another UML diagram) and it
also find further differences.
Do you have any tips / hint to avoid that ?
Regards,
Caroline
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133109 is a reply to message #133076] |
Thu, 12 February 2009 16:44  |
|
This is a multi-part message in MIME format.
--------------090406020302060902060000
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Hi Caroline,
Yes, the issue you're encountering is a lack of fonctionnalities in all
versions of EMF Compare previous to 0.9M5. Using the latest EMF Compare
builds you should now be able to properly detect cross-model
differencies (and lack of difference for that matter). Could you please
have a try with this latest version (it is backward compatible with
Eclispe 3.3 if needed)?
Don't hesitate to come back to us if this doesn't solve your issue.
Laurent Goubet
Obeo
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133133 is a reply to message #133109] |
Fri, 13 February 2009 13:31 |
| Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
Thank you for your answer.
I followed your previous hint and installed the latest version of EMF
Compare.
Here is what I did:
I installed EMF Compare 0.9.0M5 on Eclipse 3.5.0M5. I had to switch to
this version because EMF Compare needs EMF 2.5.0M5.
I created a UML model (*. uml) with UML2 3.0.0M5 and I applied the Ecore
profile : I created a Model containing a Package containing a Class
stereotyped by EClass and I saved.
I then added a property to the class, and made the comparison using the
local history. I found 815 differences and several 'diff' nodes.
Have I done something wrong? Does this issue say something to you?
I can send you my test project if it can help.
Regards
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133145 is a reply to message #133133] |
Fri, 13 February 2009 13:51 |
|
This is a multi-part message in MIME format.
--------------060302060807020109070108
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Caroline,
We have the same problem of proxy resolution when comparing with history
than we have when comparing with CVS it seems :( just tried it myself.
This falls into the scope of
https://bugs.eclipse.org/bugs/show_bug.cgi?id=199877 which we hope to
have solved for Galileo. You can workaround this by having your two
models locally for now. Resolving proxies when comparing with history
should prove less tricky than VCS proxies. I'll look into a solution for
this particular issue.
Thanks for pointing this out.
Laurent Goubet
Obeo
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133200 is a reply to message #133145] |
Tue, 17 February 2009 16:44 |
|
This is a multi-part message in MIME format.
--------------010600020801010703080100
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
FYI, a fix for this bug has been commited on HEAD and should be
available in the next builds. Subversive users will need install a new
feature "EMF Compare Subversive integration" as subversive doesn't use
the Team API as CVS does.
The behavior of Subclipse or other VCS haven't been confirmed yet, we
might very well need other such integration plugins.
Laurent Goubet
Obeo
laurent Goubet a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133478 is a reply to message #133200] |
Mon, 23 February 2009 12:23 |
| Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
Hi,
I have try with the lastest version on the CVS and it is still not
working. I have done the following test:
I have made a UML diagram with only one class. I have apply the ecore
profile and save. Then I have made a copy of this file and made a
comparaison between the two copies. It found 815 differences and several
'diff' nodes.
Do you think that this problem will be solve in the next versions? Do you
want me to open a bug on bugzilla?
Regards
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133491 is a reply to message #133478] |
Mon, 23 February 2009 12:39 |
|
This is a multi-part message in MIME format.
--------------000803030507020805010009
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Hi Caroline,
If your second file is a copy of the first and not the same compared to
history, are you sure this copy has links towards the accurate profile?
i.e if the references to the profile are relative URIs (say
.../profile/profile.ecore#//), and the copy is in another folder, there
is a chance the profile cannot be resolved.
If you think your two models are accurate and EMF Compare should be able
to resolve all proxies, feel free to open a new bugzilla, preferably
with samples to reproduce this issue.
Several "diff" nodes indicate there are more than a single file being
compared (in your case, the model and its profile makes two diff trees).
Yet using the latest version, the label should no longer be "diff" but
"x differences in <resource name>" (if you're talking about what's
displayed in the upper part of the compare editor.
Please also note that there hasn't been any builds with the latest
correction (the proxy resolution). We couldn't run a build because of a
new feature we want to offer users so that they can benefit from the
proxy resolution with subversive (the dependency won't be mandatory,
only for those who install this new feature). If you were trying with
the latest build, it is thus normal that you still have the bug. If you
were trying with the code currently on CVS HEAD ... well it should have
been fixed :(.
Regards,
Laurent Goubet
Obeo
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133503 is a reply to message #133491] |
Mon, 23 February 2009 14:58 |
| Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
I am on the CVS HEAD. In the emfdiff there are the "x differences in
<resource name>". The several diff are:
2 differences in resource My.uml
576 differences in resource ecore.profil.uml
2 differences in resource UMLPrimitiveTypes.Library.uml
138 differences in resource standard.profil.uml
816 differences in resource UML.metamodel.uml
224 differences in resource Ecore.metamodel.uml
I opened a bug with my test files attached:
https://bugs.eclipse.org/bugs/show_bug.cgi?id=265825
If you need some additionnal info don't hesitate to ask me.
Regards,
Caroline
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #133516 is a reply to message #133503] |
Mon, 23 February 2009 15:00 |
|
This is a multi-part message in MIME format.
--------------090103080704060400000406
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Would mean I don't manage to resolve proxies towards those accurately
:/. Thanks, I'll look into this ASAP.
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620649 is a reply to message #133076] |
Thu, 12 February 2009 16:44 |
|
This is a multi-part message in MIME format.
--------------090406020302060902060000
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Hi Caroline,
Yes, the issue you're encountering is a lack of fonctionnalities in all
versions of EMF Compare previous to 0.9M5. Using the latest EMF Compare
builds you should now be able to properly detect cross-model
differencies (and lack of difference for that matter). Could you please
have a try with this latest version (it is backward compatible with
Eclispe 3.3 if needed)?
Don't hesitate to come back to us if this doesn't solve your issue.
Laurent Goubet
Obeo
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620651 is a reply to message #133109] |
Fri, 13 February 2009 13:31 |
| Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
Thank you for your answer.
I followed your previous hint and installed the latest version of EMF
Compare.
Here is what I did:
I installed EMF Compare 0.9.0M5 on Eclipse 3.5.0M5. I had to switch to
this version because EMF Compare needs EMF 2.5.0M5.
I created a UML model (*. uml) with UML2 3.0.0M5 and I applied the Ecore
profile : I created a Model containing a Package containing a Class
stereotyped by EClass and I saved.
I then added a property to the class, and made the comparison using the
local history. I found 815 differences and several 'diff' nodes.
Have I done something wrong? Does this issue say something to you?
I can send you my test project if it can help.
Regards
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620652 is a reply to message #133133] |
Fri, 13 February 2009 13:51 |
|
This is a multi-part message in MIME format.
--------------060302060807020109070108
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Caroline,
We have the same problem of proxy resolution when comparing with history
than we have when comparing with CVS it seems :( just tried it myself.
This falls into the scope of
https://bugs.eclipse.org/bugs/show_bug.cgi?id=199877 which we hope to
have solved for Galileo. You can workaround this by having your two
models locally for now. Resolving proxies when comparing with history
should prove less tricky than VCS proxies. I'll look into a solution for
this particular issue.
Thanks for pointing this out.
Laurent Goubet
Obeo
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620656 is a reply to message #133145] |
Tue, 17 February 2009 16:44 |
|
This is a multi-part message in MIME format.
--------------010600020801010703080100
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
FYI, a fix for this bug has been commited on HEAD and should be
available in the next builds. Subversive users will need install a new
feature "EMF Compare Subversive integration" as subversive doesn't use
the Team API as CVS does.
The behavior of Subclipse or other VCS haven't been confirmed yet, we
might very well need other such integration plugins.
Laurent Goubet
Obeo
laurent Goubet a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620678 is a reply to message #133200] |
Mon, 23 February 2009 12:23 |
| Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
Hi,
I have try with the lastest version on the CVS and it is still not
working. I have done the following test:
I have made a UML diagram with only one class. I have apply the ecore
profile and save. Then I have made a copy of this file and made a
comparaison between the two copies. It found 815 differences and several
'diff' nodes.
Do you think that this problem will be solve in the next versions? Do you
want me to open a bug on bugzilla?
Regards
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620679 is a reply to message #133478] |
Mon, 23 February 2009 12:39 |
|
This is a multi-part message in MIME format.
--------------000803030507020805010009
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Hi Caroline,
If your second file is a copy of the first and not the same compared to
history, are you sure this copy has links towards the accurate profile?
i.e if the references to the profile are relative URIs (say
.../profile/profile.ecore#//), and the copy is in another folder, there
is a chance the profile cannot be resolved.
If you think your two models are accurate and EMF Compare should be able
to resolve all proxies, feel free to open a new bugzilla, preferably
with samples to reproduce this issue.
Several "diff" nodes indicate there are more than a single file being
compared (in your case, the model and its profile makes two diff trees).
Yet using the latest version, the label should no longer be "diff" but
"x differences in <resource name>" (if you're talking about what's
displayed in the upper part of the compare editor.
Please also note that there hasn't been any builds with the latest
correction (the proxy resolution). We couldn't run a build because of a
new feature we want to offer users so that they can benefit from the
proxy resolution with subversive (the dependency won't be mandatory,
only for those who install this new feature). If you were trying with
the latest build, it is thus normal that you still have the bug. If you
were trying with the code currently on CVS HEAD ... well it should have
been fixed :(.
Regards,
Laurent Goubet
Obeo
Caroline Bourdeu d'Aguerre a
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620680 is a reply to message #133491] |
Mon, 23 February 2009 14:58 |
| Eclipse User |
|
|
|
Originally posted by: cbda.free.fr
I am on the CVS HEAD. In the emfdiff there are the "x differences in
<resource name>". The several diff are:
2 differences in resource My.uml
576 differences in resource ecore.profil.uml
2 differences in resource UMLPrimitiveTypes.Library.uml
138 differences in resource standard.profil.uml
816 differences in resource UML.metamodel.uml
224 differences in resource Ecore.metamodel.uml
I opened a bug with my test files attached:
https://bugs.eclipse.org/bugs/show_bug.cgi?id=265825
If you need some additionnal info don't hesitate to ask me.
Regards,
Caroline
|
|
|
| Re: [EMFT Compare] Additionnal ressource and compare [message #620681 is a reply to message #133503] |
Mon, 23 February 2009 15:00 |
|
This is a multi-part message in MIME format.
--------------090103080704060400000406
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 8bit
Would mean I don't manage to resolve proxies towards those accurately
:/. Thanks, I'll look into this ASAP.
Caroline Bourdeu d'Aguerre a
|
|
|
Powered by
FUDForum. Page generated in 0.04664 seconds
] 
Search
Help
Register
Login
Home
