Simon Laws wrote:
[snip]
Looking back at the spec we are maybe missing the notion of a contribution
being installed (p67 of the assembly model) vs it just being contributedto
the domain. So we should revisit what we think the lifecycle of a
contribution is. Something like...
created (by some assembler)
contributed (to a domain)
installed (resolved w.r.t it's dependencies)
deployed (to a node where a composite requires the installed contribution
artifacts in order to run)
Does that make sense?
I'm almost on the same page, except that I'd change "installed" to
"validated".
Here's an analogy that should help illustrate the idea:
- as an app developer I work with an Eclipse workspace containing projects
- as I add/remove/change projects I have the option to validate and/or
build my workspace
- as I start a JVM to run a program out of a project I ensure that the
project dependencies are there and the program's class compiles OK.
Replace developer by administrator, workspace by domain, project by
contribution, JVM by node and program by composite and you'll get close
to my view.
Hope this helps.
--
Jean-Sebastien
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
