Let's do a thought experiment. I've decided to use
Java 11 in EMF and have decided to change the BREEs as
such. Is anyone else's opinion binding? Maybe the
Modeling PMC's opinion would be, but I'm the lead and
clearly I agree with myself. I don't think the
Planning Council's opinion is binding is it? I see no
rule stating that I'm not allowed to use Java 11 for my
project and my right to determine my project's course is
very important to me. In any case, I'm on the Planning
Council, though with spotty attendance, and I could
argue that they have no jurisdiction. I would also
think that the Eclipse Packaging Project's opinion is
not the least bit relevant to me personally nor to my
project. After all, I have no binding opinion there,
and if an opinion isn't binding, it follows that it has
no real weight and is of little value. So in the end,
from my project-centric view, everyone should consider
themselves lucky to have EMF at all.
What problems would this cause? Nothing serious
technically. Just change the required Java version for
all products and your done. Get over it already.
I fully agree with everything you said! Whoever is
interested in having a say on EMF future should be investing
in EMF (either directly hiring people or paying others to do
it for them) in order to have a binding vote on EMF future.
All of us the rest should be happy that we have it (or any
other project!) as long as we don't invest in it somehow.
Yes, it boils down to "beggars can't be choosers." Beggars get
what they're given and can are hardly be in a position to make
demands.
So we can also be lucky that Mickael and team provide Corrosion
and the Rust package and he should have the freedom do guide it as
he see fit.
I suppose the Planning Council could decide that some
old version EMF should be on the train instead, and the
EPP can then just use that old version. But then you'd
all be on your own, without new releases nor bug fixes,
ever.
Clearly it would be better if I were sensitive to the
community's opinions and needs rather than focused so
much on what is and is not binding.
Sensitivity is not helping much in attracting new people.
IMHO it's counter productive to this for 2 reasons:
1. Individuals are not jumping in to help because of
numerous requests, rules, demands and etc. coming from
everywhere which quite often mean a whole lot of additional
work for the individual for pretty much zero benefit for
him/her.
This is an excellent counter point! I invest significant effort
keeping EMF compatible with really old things, and I even test
that EMF actually works on such really old things:
But why? What's benefit to me? Certainly more people can use
the latest EMF in more places, but I imagine that also makes
everyone very complacent that their high expectations are always
satisfied.
2. Corporations are not jumping in to help because they
know they can demand, request, try to enforce rules and that
this works most of the time even if they reduce their
investment or not invest at all.
Yes, corporations with pockets so deep you could put a million Euros
in there and never find it again.
So the real question is what is community?
Everyone.
Is it every entity that makes some use and don't
contribute back or the level of contribution is very
unproportional?
Yes, even the beggars are part of the community.
This is clearly not my understanding of "community".
Certainly the greater community could benefit from being a greater
community! Being sensitive to the beggars and freeloaders is not
necessarily a good thing. It just encourages them. It's a point
well taken.
_______________________________
The issue of arguments for or against a higher BREE?
The arguments for are clear. We get to use the latest
and greatest Java language features and most current
Java library classes and methods. That's strongly
compelling when considered in isolation of all other
considerations.
The hint about needing Java 8 on bottom sidebar to the
right would need to mention more specifically the need
for Java 11 for some of the packages. No big deal, but
would need attention.
Users launching a newly unpackaged package will be
informed that they need Java 11 (or the installer will
inform them, though if EMF changes the installer itself
would need Java 11 not just Java 8). No big deal; the
user will hunt down the right thing from somewhere. A
little worse is that users who install/update into an
existing installation will find their installation
broken if it was EMF involved, because nothing in the
IDE will start without EMF starting, or will find parts
of the installation not working for mysterious reasons.
I suppose also no big deal when it's some leaf component
involved.
None of these consequences are ideal though and
unfortunately we have no statistics to indicate how much
more likely a user is to have Java 8 already installed
versus Java 11 or higher already installed. But clearly
setting the bar higher will leave more of the community
below the bar while leaving the bar where it is now is
most definitely less disruptive.
So how would moving Eclipse IDE for Rust moving to Java
11 leave more of the community out?
It's just an issue of how likely is it that people will have
problems because they need to install a newer Java version or will
even realize they have problems because they don't have a newer Java
version.
And of which community? Rust one ?- I doubt if anyone
there cares about Java version being used at all (btw, this
is probably true for C/C++ too).
Mickael will understand this community best and I have not argued
they he should not move to Java 11. If that's best for his
development and his community he should be free to do what he feels
is best. I've merely pointed out some things to consider while
making that decision.
Eclipse IDE community? - How much anyone in the Planning
Council or any other project care about Rust tooling?
Indeed, it's a pretty tiny part. But a cool part.
Corrosion uses RLS for all the smartness and the idea is
to reduce testing matrix and ease support for that project.
If we want to get into theoretical discussions it's one
thing but let's at least try to prove that these theories
apply to the case in question.
Yes, although this thread started, I've seen Gunnar suggest that all
packages move to Java 11 for 2020-06. I'm not at all convinced that
that's such a great idea.
On 02.03.2020 07:16, Aleksandar Kurtakov wrote:
On Sun, Mar 1,
2020 at 8:28 PM Mickael Istria <mistria@xxxxxxxxxx>
wrote:
No,
it is not OK for the same arguments
mentioned before.
Can you please remind those arguments? I'm
curious about which ones apply to typical
Corrosion target audience.
Also -and without any offense but more to
clarify who has influence/authority on EPP
project as it's facing a transition-, I'm
curious about whether your opinion as Planning
Council member but non-EPP committers is to be
considered as binding. WDYT?
Planning Council has no bigger right over EPP
(compared to any other project) to say whether EPP
project allows some of its deliverables to require
Java 11. It was specifically discussed at the last
Planning council and we agreed on that https://wiki.eclipse.org/Planning_Council/2020-02-05#Notes
. This is a decision which should be driven by the
EPP project committers together with the
not-yet-elected lead.