http://www.eclipse.org/emf/facet/serialization/0.1.incubation wasn't renamed

[eclipse-modisco-bot]

| --- | --- |
| Bugzilla Link | 559269 |
| Status | NEW |
| Importance | P3 normal |
| Reported | Jan 16, 2020 11:39 EDT |
| Modified | Mar 06, 2020 12:32 EDT |
| See also | 559205, Gerrit change https://git.eclipse.org/r/158747, Gerrit change https://git.eclipse.org/r/158749, Gerrit change https://git.eclipse.org/r/158748, Gerrit change https://git.eclipse.org/r/158746, Git commit 66fcaf17, Git commit c483b9b9, Git commit 6e4b271d |
| Reporter | Ed Willink |

Description

Bug 599205 identifies that

http://www.eclipse.org/emf/facet/serialization/0.1.incubation

is not registered and observes that emf/facet should have been renamed.

This bug rename all emf/facet.

[eclipse-modisco-bot]

By Ed Willink on Jan 16, 2020 11:46

Also http://www.eclipse.org/emf/facet/efacet/0.2.incubation ...

and a puzzling number of *.custom references

[eclipse-modisco-bot]

By Fabien Giquel on Mar 03, 2020 09:28

Some users (like me) may have reasons to install both new Modisco1.5 and old released EMFFacet1.3 : then the problem becomes critical because the both register EPackages implem with same nsURI.

I will make the following fixes :

• replace all "emf/facet" to "modisco/facet" in nsURI from various facet .ecore , and related generated .java
• replace all related "emf/facet" to "modisco/facet" in .custom
• while we're at it, fix relative references in .ecore/.genmodel to old "emf.facet" plugins, and baseDirectory in .genmodel

I think such nsURI evolution is only adjustment/fix in big 1.5.0 evolution scope, and so we should not increment major or minor modisco version...?

[eclipse-modisco-bot]

By Ed Willink on Mar 03, 2020 09:36

Ok. That absolves my conscience. The nsURI is relatively easy, but the number of marginal, e.g. documentation hits, on "emf/facet" depressed me.

Please do it today so that we can contribute it to M1 tomorrow.

[eclipse-modisco-bot]

Mar 03, 2020 11:42

New Gerrit change created: https://git.eclipse.org/r/158747

[eclipse-modisco-bot]

Mar 03, 2020 11:42

New Gerrit change created: https://git.eclipse.org/r/158749

[eclipse-modisco-bot]

Mar 03, 2020 11:42

New Gerrit change created: https://git.eclipse.org/r/158748

[eclipse-modisco-bot]

Mar 03, 2020 11:42

New Gerrit change created: https://git.eclipse.org/r/158746

[eclipse-modisco-bot]

Mar 03, 2020 15:29

Gerrit change https://git.eclipse.org/r/158746 was merged to [master].
Commit: http://git.eclipse.org/c/modisco/org.eclipse.modisco.git/commit/?id=66fcaf1738c79aa2c1d14c7db973d8e719e724f3

[eclipse-modisco-bot]

Mar 03, 2020 15:30

Gerrit change https://git.eclipse.org/r/158747 was merged to [master].
Commit: http://git.eclipse.org/c/modisco/org.eclipse.modisco.git/commit/?id=c483b9b9264978c2bff3dc48aeda4167b645bd8a

[eclipse-modisco-bot]

Mar 03, 2020 16:18

Gerrit change https://git.eclipse.org/r/158748 was merged to [master].
Commit: http://git.eclipse.org/c/modisco/org.eclipse.modisco.git/commit/?id=6e4b271d3b02c18585e72570df03a691c2541821

[eclipse-modisco-bot]

By Ed Willink on Mar 04, 2020 03:47

(In reply to Fabien Giquel from comment #2)

Some users (like me) may have reasons to install both new Modisco1.5 and old
released EMFFacet1.3

This is an interesting, but not necessary use case.

Hard-nosed View:

Modern EMF provides no ability to co-exist with the pre-generics EMF. If I want to use Ecore Tools, the only version that is useable for me is the penultimate 1.x version. For this I have to use Helios.

While EMF Facet is not yet quite dead, its termination review is overdue and unopposed, indeed advocated by you. If someone wants EMF Facet they should have spoken up, avoided the GIT being archived, and adjusted the master branch to co-exist with the live MoDisco project.

For the emf/facet nsURI there could be some debate. For the clashing GMT MoDisco nsURI it's clear, it belongs to MoDisco. Sorry about EMF Facet. Tough.

More helpful view:

When I saw the variety of 0.x versions in nsURIs, I considered changing them to 1.x since a mature 1.x project should not have 0.x components. But I was reluctant to make such a potentially big change when there was actually no underlying change in content or quality. It might appear that this could have avoided your EMF Facet clash. But actually I would have added a legacy duplicate so that any use of the old nsURI would have automatically used the new one. Clash preserved / re-inforced.

Way forward:

It is too late to do something contentious for 2020-03.

I could imagine another pseudo-major version change to 1.6.0 for 2020-06 in which residual emf/facet was fixed and maybe 0.x => 1.0.

What I do not know is to what extent these nsURI are transient, i.e. a user writes then reads so a spelling change is irrelevant, or whether there are some users with legacy emf/facet/0.1 files that will be broken unless a legacy nsURI registration is preserved.

[eclipse-modisco-bot]

By Ed Willink on Mar 06, 2020 06:51

(In reply to Ed Willink from comment #11)

It is too late to do something contentious for 2020-03.

I'm not clear how this contentious change has made it to master. Non-trivial changes should surely happen on a branch and only once approved be rebased master? That's why we have branches.

I have reverted the 3 Bug 559269 commits awaiting an agreement as to if/how/when to proceed.

I have left the Bug 553310 commit affecting only .checkstyle which seems to affect only tools that users/myself have limited/no familiarity with.