Eclipse Community Forums: Compare » Papyrus model comparison results in a lot of non-existing differences

Home » Modeling » Compare » Papyrus model comparison results in a lot of non-existing differences

Papyrus model comparison results in a lot of non-existing differences [message #1749442]

Tue, 06 December 2016 13:50

Eclipse User

When I try to compare a workspace version of a simple Papyrus model with an earlier version stored in a local git-repo using "Compare With->Commit..." from the context menu the resulting differences in the comparison editor indicates a lot of differences even if there are only one element that have been moved slightly in a diagram.
The right side of the comparison editor indicates that something have not been resolved correctly as elements are presented staring with "Proxy:Platform resource..." instead of the actual name of the element. See attached screenshot
I can also mention that comparison of two versions of the same model in the workspace works as expected. The problem only occur when comparing a model in the workspace with a commited model in Git.

I am running Papyrus 2.0.1 on Neon.1 on Windows 64 and have installed EMF Compare 3.3 with support for UML2, GMF and Papyrus, see attached screenshot.

I have selected the "Recursive Model Merge Strategy" and also "Pre-merge models when a real conflict is detected" in the preferences.

Have I missed something in the installation around EMF Compare/Git support or are there any more preference settings to tune?

Attachment: Screenshot EMF Compare comparison.png
(Size: 78.27KB, Downloaded 315 times)
Attachment: Screenshot EMF Compare installation.png
(Size: 80.55KB, Downloaded 306 times)

Re: Papyrus model comparison results in a lot of non-existing differences [message #1749476 is a reply to message #1749442]

Wed, 07 December 2016 03:35

Eclipse User

Hi Thomas,

it looks like the Test_Model.uml could not be resolved on the right-hand side, for some reason. It does still exist in the commit you compare your workspace to, right? Was it maybe renamed or moved?

Do you use the integration builds from collaborative-modeling.org? The version numbers indicate that you are not, but I didn't double-check.

Thanks and best wishes,

Philip

--
Philip Langer

Get professional Eclipse developer support:
http://eclipsesource.com/en/services/developer-support/

Re: Papyrus model comparison results in a lot of non-existing differences [message #1749484 is a reply to message #1749476]

Wed, 07 December 2016 04:28

Eclipse User

Hi Philip,

I have not renamed nor moved the Test_Model.uml at all.
Just created the model and shared it in a new local Git repo, made a first commit and then tried to make some minor changes (for example moving a comment in the diagram). saved the changes and did a "Compare With" with the commited version in Git.

I can say I have tried the same scenario on other customer-specific models in other installed Papyrus/EMF Compare Environments with the same result.

The EMF Compare plugins was installed from the official 3.3 update site (http://download.eclipse.org/modeling/emf/compare/updates/releases/3.3/) all parts was selected.
I thought that was the recommended version to use.

But I will try with a newer integration build as well.

Re: Papyrus model comparison results in a lot of non-existing differences [message #1749492 is a reply to message #1749484]

Wed, 07 December 2016 05:51

Eclipse User

I finally successed in getting the comparison to work correctly

I totally reinstalled Papyrus 2.0.1 RCP and on top of that the components from the two repositories belonging to the integration build at collaborative-modeling.org.
I also had to switch to a brand new workspace and reconfiguring the recommend preference settings.

I must say that it is a bit confusing that the official 3.3 build (which is newer than the integration build) does not work, but that the "older" does. Trying to "upgrade" from the official 3.3 build to the integration build will therefor fail.

Thanks for the support on this issue Philip

Re: Papyrus model comparison results in a lot of non-existing differences [message #1749504 is a reply to message #1749492]

Wed, 07 December 2016 08:25

Eclipse User

Thomas Wiman wrote on Wed, 07 December 2016 10:51

I must say that it is a bit confusing that the official 3.3 build (which is newer than the integration build) does not work, but that the "older" does. Trying to "upgrade" from the official 3.3 build to the integration build will therefor fail.

Yes, we have had issues before when the integration builds of the custom builds of EGit and EMF Compare have been "lagging behind" the official releases of EGit and EMF Compare. Keep in mind that the custom builds provided at www.collaborative-modeling.org contains specific handling of logical model merge which unfortunately have not completely made its way into the official EGit and EMF Compare releases yet. I do know that one of the "missing pieces" for EGit got incorporated yesterday on the master branch in the EGit repo (which I assume will be included in upcoming EGit 4.6 release).

But I guess that Philip can fill in here regarding the details.

/Peter Cigéhn

Re: Papyrus model comparison results in a lot of non-existing differences [message #1749530 is a reply to message #1749504]

Wed, 07 December 2016 13:50

Eclipse User

Hi Thomas,
hi Peter,

you are indeed right, it is confusing with the custom builds. We try to communicate that as clearly as possible and work on resolving this asap. In fact, Laurent D. (Obeo) has succeeded in getting the last most crucial commit merged in EGit today. So things are looking good for the next release, but we'll have to see and continue testing, if now everything crucial is included in EGit.

As it turns out, the issue you are having in your comparison (as shown in your screenshot), is because the comparison has been started from the project and not from the model file. If the comparison is started from the model, the comparison works well. I could reproduce this issue and took the liberty to create a bug report (bug 508855). After looking into this bug, I found the reason for it and sketched a solution, which I'll push for review and testing soon. I'll have to check if my solution has any side-effects first. Feel free to cc yourself in bug 508855, if you want to be notified on the progress.

Thanks for reporting this issue!

Best wishes,

Philip

--
Philip Langer

Get professional Eclipse developer support:
http://eclipsesource.com/en/services/developer-support/

[Updated on: Wed, 07 December 2016 14:21] by Moderator

Previous Topic:	Record changes
Next Topic:	EMFCompare

Goto Forum:

-=] Back to Top [=-

Current Time: Mon Jul 14 04:38:00 EDT 2025

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter