Bug 518289 - Implicit dependency between 2 migration contributions DRepInDViewToRootObjectsAndWithDRepDescRepPathMigrationParticipant and NoteShapeDefaultLabelAlignmentMigrationParticipant
Summary: Implicit dependency between 2 migration contributions DRepInDViewToRootObject...
Status: NEW
Alias: None
Product: Sirius
Classification: Modeling
Component: Core (show other bugs)
Version: 5.0.0   Edit
Hardware: PC Windows 7
: P3 normal (vote)
Target Milestone: Next   Edit
Assignee: Project inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords: triaged
Depends on: 518644
Blocks: 529263
  Show dependency tree
 
Reported: 2017-06-15 05:16 EDT by Minh Tu Ton That CLA
Modified: 2018-09-05 08:32 EDT (History)
2 users (show)

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Minh Tu Ton That CLA 2017-06-15 05:16:23 EDT 
Hi,

During the migration of aird, migration contributions are called (see org.eclipse.sirius.business.internal.migration.AbstractSiriusMigrationService.postLoad(XMLResource, String)). I've just been aware that there must be a dependence between the contribution NoteShapeDefaultLabelAlignmentMigrationParticipant and the contribution DRepInDViewToRootObjectsAndWithDRepDescRepPathMigrationParticipant. If the latter is not called before the former, the migration of note alignment is not performed. 
While running a test with Tycho, the order of plugin loading is org.eclipse.sirius.diagram.ui then org.eclipse.sirius which makes the contribution NoteShapeDefaultLabelAlignmentMigrationParticipant precede the the contribution DRepInDViewToRootObjectsAndWithDRepDescRepPathMigrationParticipant in the list.
Is it a must to assure the dependency between migration contributions somehow ?
Comment 1 Pierre-Charles David CLA 2017-06-15 10:54:35 EDT 
I'll try to test later, but it looks like we should indeed guarantee the particpants are executed in a deterministic and chronological order. https://git.eclipse.org/r/#/c/94974/ may be enough, but further investigation is needed.

Is the problem systematic in you case when you run your tests under Tycho?
Comment 2 Minh Tu Ton That CLA 2017-06-15 11:30:20 EDT 
Yes, it's systematic. Our current workaround is to ensure the dependency order of sirius plugins in the test plugin.
Comment 3 Pierre-Charles David CLA 2017-12-28 03:40:58 EST 
There are several other related issues regarding migration participants and their coordination (#519596, #506226, #518644). We'll try to have a more global look on these for 6.0.