We had a discussion about this in our
last PMC call. We talked about the following options:
1) Migrate tools into a new project
2) Migrate tools into PDE
3) Migrate tools into Platform UI
Option 1) is always a possibility. There
is some added overhead with each new project, such as committer elections
and various other bits of Eclipse process. In general if there is an existing
project that is a good fit I would recommend that over the work of creating
an indefinitely maintaining a new project.
Option 2) makes sense on a conceptual
level because PDE is the home of all tooling specific to the Eclipse platform
runtime. However there is absolutely no connection between these tools
and the existing PDE code base, and no overlap between committers. So it
"fits the category" but otherwise has no common ground with the
contents of that project. Also, once modularity comes to the Java language,
we will likely see PDE align more closely with JDT, and the e4 tooling
doesn't fit with that.
Option 3) is compelling because there
is a strong overlap between current committers on both tools and runtime,
and of course close relationship between the tooling and runtime code -
when one has significant changes the other likely needs to react to it.
After some discussion, all members of the PMC are in favor of this option
and this is what we recommend. This would be implemented by creating a
new Git repository under Platform UI project to host the tools, and then
elect all active contributors on the graduating tooling into Platform UI.
It would initially be a separate feature that is available in the project
repository that is installed separately (like Eclipse Releng Tools, for
example). This would immediately accomplish the goal of making it easy
for end users to install into Eclipse Mars and beyond. In the future it
could be added to EPP packages where that makes sense (such as the RCP
So Option 3) is the current PMC recommendation,
but if the e4 tools contributors want to take it in a different direction,
such as a new project, we are happy to talk about it. What do you think?
Lars Vogel <lars.vogel@xxxxxxxxx>
08/27/2014 11:38 AM
Restructuring review for the "e4 tools" Git project migrating
to its own project
thanks. Call sounds like a good idea.
Best regards, Lars
2014-08-27 14:57 GMT+02:00 John Arthorne <John_Arthorne@xxxxxxxxxx>:
We had to cancel our PMC call today
because several of us were out or unavailable, so we'll need to talk about
this next week. I have to say I always thought of PDE as the eventual home
for all those tools. Note that being under the same project does not necessarily
mean they have to be included in the same feature as the basic plugin tooling.
It could be built as one or more separately installable features that live
under the same project. They could even live in a separate Git repository
from the other PDE code. The only tangible difference with a new project
is that it has a distinct committer group. But in the long term I'm not
sure there will be a large enough committer base for that to be needed.
Would it be worth setting up a call to talk about the different options
e4 tools provide the tools to create and work with Eclipse 4 based applications,
IDE's as well as RCP applications.
Parts of the tools are planned to get integrated into PDE, i.e. the e4
project wizard. But others, like the application model editor and the "spies"
are currently not planned to be integrated to PDE as they (in parts) depend
on extensions of the EMF framework, like the EMF undo / redo support.
The e4 tools would like to get included into the "official" Mars
release. Our users complain that they have to seek a different update site
to get the tools integrated.