Re: [mdt-papyrus.dev] About tests

[LETAVERNIER Camille <Camille.LETAVERNIER@xxxxxx>]

Ø  (everybody does run the tests locally before they push changes, right?  ;-)

Sure :) Remember this issue about results being different on Tycho and JDT ? :)

 

SVG images seem to be failing on the Hudson/Tycho build for some reason. They caused a lot of indirect test failures (Took me 6 weeks to figure it out... I guess the over-spammed error log in the test build didn’t help). But these tests were mostly fine, when executed locally.

 

Camille

 

De : mdt-papyrus.dev-bounces@xxxxxxxxxxx [mailto:mdt-papyrus.dev-bounces@xxxxxxxxxxx] De la part de Christian W. Damus
Envoyé : jeudi 12 février 2015 14:21
À : Papyrus Project list
Objet : Re: [mdt-papyrus.dev] About tests

 

Hi, Camille,

 

I had fixed the ExtendedTypes tests after they broke on adoption of GMF Mars M5 but they regressed when the element-types configurations model branch was merged.  I haven’t looked into that (they passed on that branch the last I recalled).

 

The decorator-model control-mode test failures are tracked in a bugzilla (don’t have the reference) and Gabriel mentioned he would be looking into them.

 

It probably needn’t be said, but it would be really nice to get a green build so that committers can rely on the tests to find new regressions.  As it stands, it is hard to run the tests locally to check for regressions and compare the results against many failures in the build.  (everybody does run the tests locally before they push changes, right?  ;-)    Let’s keep it green.

 

Cheers,

 

Christian

 

 

On Thu, Feb 12, 2015 at 8:03 AM, LETAVERNIER Camille <Camille.LETAVERNIER@xxxxxx> wrote:

Hi all,

 

 

The tests have been failing for a while now; mostly caused by the important diagam refactorings on Mars (Figures in M4, Element Types in M5).

Most M4 regressions have now been fixed. All regressions on Luna have been fixed as well (Except one test in the Extra components, related to the CDO/ControlMode integration)

 

We now have 52 remaining failures on Mars, mainly related to the M5 refactoring (So nothing critical here; everything’s pretty much expected). I’ve done what I could to fix some failures, and cleaned up the build log from tons of redundant exceptions (Not everything has been removed, but most of them)

 

About the remaining failures:

 

-          BDD Drag & Drop tests are failing (14 direct failures, + 2 failures on the delete test because the test relies on D&D to initialize the model)

o   16 Failures

-          SysML Associations have a different semantics from the UML Associations. It seems their specific behavior has been overridden; they now behave like UML associations

o   8 Failures

-          InteractionOverview Diagram: it extends the ActivityDiagram, which has been migrated to the new Element Types, but the InteractionOverview itself has not been migrated

o   13 Failures

-          CompositeDiagram: Failures are related to the refactoring in M4. I’ve been investigating on these failures back then; and couldn’t figure what’s wrong (The behavior is fine at runtime but fails in the tests). I’ll investigate more on these ones later.

o   4 Failures

-          ExtendedTypes: Are these tests deprecated, since we now use the new Element Type metamodel? Or are they still expected to work?

o   7 Failures

-          DecoratorModel/ControlMode: Not investigated

o   3 Failures

-          ClassDiagram/Containment Link: I think the EditPart/Figure has been simplified, so the test may simply need to be updated

o   1 Failure

 

 

Regards,
Camille