De-facto current operational requirements are at least Java 5 and
Eclipse 3.2. I think Alex or Nick can add more details there.
Let talk road map in separate discussion.
For JSR-291, I think we better follow Equinox position in that
subject, and, frankly, I cannot speculate what it will be. Anyway, all
this relationship between Java (Sun), Eclipse (IBM?) and OSGi are big
unknown for me. Do anybody have strong opinion toward JSR-291?
Best,
Igor.
Wendell Beckwith wrote:
This is a discussion I was hoping to have, however I wanted the final
proposal to see if things were different before opening a new can of
worms. Nonetheless, the minimum operational environment for the ecp
needs to be specified. Is it java 1.4 or java 5? I saw the email
that Exadel had added more components. Perhaps those additions (I
admit I haven't looked at them) are orthogonal and won't be affected
by changing the core framework except maybe at the activator and/or
bundle listeners.
But if that is not the case then we would be well served to focus on
specifying and then modifying/writing the core framework for java 1.4
or 5. Personally I'm all for java 5 because I like most of the
features in it, but general generic exposure to end users simply
isn't one of them.
Another issue is on road maps for milestones and releases. I don't
feel we need to wait for project creation review to specify what we
think would be good additions for the ecp.
And my final issue is JSR-291 and what that means for the ecp and
well as the equinox projects. I believe our focus should be to
follow this spec. as it works its way through the JCP. Key things
that will affect us is whether equinox will adopt it or do they plan
to remain a relatively "pure" OSGi implementation.
Igor, Alex thoughts?
Wb
Jeff McAffer wrote:
Yes! Actually, we avoid anything in Java 5 unless absolutely
required. The
problem is that it reduces the number of places that people can use the
plugins. So unless the gain is significant, the loss is not
outweighed.
Jeff
"Wendell Beckwith" <wbeckwith@xxxxxxxxx> wrote in message
news:e2lip9$kev$1@xxxxxxxxxxxxxxxxxxxx
Early last week and today I got time to dig into the core plugin
and in
an effort to understand what you all had done I started a new
plugin and
started rewriting it. There are serveral inconsistencies where some
interfaces begun with an 'I' and some didn't among other things. And
from having reimplemented nearly half the core plugin I truly question
the utility of using generics in the core. If you review the pdfs @
http://www.oreillynet.com/pub/wlg/7202, you can see the almost utility
of generics. Our current code is java 5 so we use java 5 features,
but
generics really complicates the code more than it solves. For my team
the rule is that generics can be used in our own code internally
but no
generics are allowed in API classes that end users would see.
Wb