This will take some time, just to think
through, and discuss with others.
I suggest we discuss at our next "first
Wednesday meeting" (April 2) to decide what Planning Council's position
is on this.
Thanks for raising it here.
In the mean time, I will leave a few
thoughts in the form of questions:
Can we clearly state "what problem
we are trying to solve"? Can we clearly state what benefits this would
More specifically, is this to benefit
adopters? Is its purpose to get more feedback before Luna is released?
How much of the benefits could be obtained
by simply having a wiki page that explained "who's provided Kepler
SR2 patches" (and, providing a list of "patch URLS")?
My intuition is that your two month
estimate is low, and that the best time to do a high quality SR3 would
be after Luna ... and don't forget, any "post Kepler" work done
now, will mean less work done for Luna (and Mars) ... and if that is the
case, is it valid to have an SR after Luna? or does that put it into "LTS
Would we be setting expectations too
high? I know that even now, what is in the "JDT/PDE patch for Kepler
SR2" is not as functional as what is in Luna I-builds (in some minor
ways, right now, but would expect that to grow) so would the community
think a Kepler SR3 have the same Java 8 support as Luna? And then be disappointed
to learn that wasn't the case?
I have a more open mind than my questions
might suggest ... but these are sincere questions I've not seen addressed
(admittedly, some may have been, and I skim read too much).
And, I won't speak for Markus, but think
you can take it as a good starting assumption that I won't be involved
with the "aggregation work" -- I get little enough sleep as it
Ian Bull <irbull@xxxxxxxxxxxxxxxxx>
03/23/2014 05:04 PM
Kepler J8 Release
There has been a discussion on the CrossProject ML about
a Kepler Java 8 release. If there is some, non-null segment of our community
that this would help, I don't think we should completely ignore this request.
However, I don't want to overstep my bounds, so I thought I would raise
this here first.
Dani gave a good summary of why we should not "simply"
(big air quotes there) release an SR3. Because of the changes required
to the JDT, there is a good chance that this release is less stable than
SR2, and while we have been discussing the reality that SRs are not more
stable, the JDT + Platform is different. If this idea did go forward, I
think a separate release 'Kepler J8' would be required.
Here is a draft response:
If we are serious about moving forward with a J8 release
for Eclipse, someone in the community will need to put a concrete proposal
together. The proposal should cover:
assume this will also include a SimRepo as well as packages. Most of this
repo should just be Kepler SR2, but we need to decide what updated plugins
should be included (JDT for sure, maybe WTP, M2E, etc...). We need a list
of updated plugins, and a +1 from the projects that they want their plugins
some of these components have not been 'released' yet. Release reviews
will need to be organized (let's not assume the projects will do this).
probably don't need to spin every package (CDT?, Modeling?), so we need
to decide which packages to build.
list of resources, who will coordinate this, aggregate the repository and
build the packages. Let's not assume David and Markus will have the time
to do this.
QA / Sign-off process. Who is going to QA the packages, and what platforms
will we support.
If someone puts a proposal together that addresses these
points (and are capable of providing the resources to pull this off) would
we be ok with it? It would also be good to put a date on this. If this
work will take 2 months, bringing us to late May, is this still valuable?
Thoughts? Is this worth discussing? Or should we just
stall long enough and let Luna leave the station ;).