Mickael,
Comments below.
On 10.12.2019 10:05, Mickael Istria
wrote:
Hi Ed,
Thanks for bringing this.
I was actually waiting for the best opportunity to start
discussing a very related topic, but from a maintenance POV
not a user one, and I think this is a good one:
*What is the value and quality of SimRel? Is it worth it's
cost?*
As you're highlighting, the actual SimRel doesn't result in
good "common" toolset. Installing whole SimRel leads to a
failing or non-usable IDE. So is this non-functional thing
worth it?
I'd say yes, because not doing this just moves the problems from
being centrally noticed to being even more widely distributed
instead.
EPP packages behave fine, they're tested and delivered to
the target audience in a good enough shape. Marketplace is now
the main and best entry-point for many features and fits much
better into typical user workflows
Stay tuned for my next email about marketplace. :-P
Does SimRel bring anything really good beyond EPP packages?
I think so. But I also suspect we have accumulated much baggage on
the train...
Anything good enough to be worth the maintenance effort?
In the end we still need a train-like thing for EPP. We could put
less on the train. And I think the cost is not so much maintaining
the train but rather producing the things on the train. I feel
that most train-level things could be more stream-lined and further
automated. In principle, why could we build staging packages
daily? What all needs to be done that can't be fully automated?
Is it an interesting exercise to expect all Eclipse
Platform plugins in the Eclipse Foundation to work properly
together as the same application?
This is a fundamental aspect of traditional Eclipse platform stack
of technology. That you can add features as needed for the
development purposes.
Who is actually likely to invest in making their specific
plugins fit better with all other Eclipse plugins if the
typical audience doesn't care?
I would not assume that the end users don't care. Don't most of us
and our end-users add our favorite tools to some base installation?
My opinion on this is that SimRel is nowadays a giant hard
to maintain inconsistent effort, and that EPP and Marketplace
have superseded it and made it not so profitable anymore.
I don't think it's so hard to maintain. It's the content that's the
major effort. But if we don't bother to look at how well content
will integrate and interoperate, the problem won't go away. It's
more likely to get worse, and as far as the user is concerned all
that is "Eclipse's fault".
I would suggest that we just abandon SimRel in favor of EPP
building packages and a p2 repository containing all the
dependencies for these packages; and other plugins should just
target Marketplace as entry point, until they reach a good
enough quality and value to get into some EPP package.
Personally I think that's a bad idea. The train repo is after all
the input to the EPP packaging build, so at best we could reduce the
train to what's actually used by the packages. Or so it seems to
me anyway.
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/cross-project-issues-dev
|