On Mon, Jul 22, 2013 at 04:30:32PM +0200, Miloslav Trmač wrote:
How do I package and ship an application that depends on two or more
stacks (e.g. uses a database, has a web frontend and perhaps also a
local GUI) if the stacks the application depends on can have
overlapping and therefore probably (necessarily?) conflicting
software, different lifecycles and packaging methods?
I'm not suggesting a crazy free-for-all. If you have an application which
depends on multiple stacks, you would want to select one technology for all
of those stacks -- OpenShift cartridges, or Software Collections, or Fedora
Formulas. If you want to use a cartridge for one and a software collection
for the other, I don't think there's any expectation that this would work.
This is also why I had "environments" as a higher ring in earlier versions
of the slides. When you're making your development or package-and-ship
decision, you want to be thinking about the environment in which it will run
and then the implementation technology decision comes from that.
> This is a special bubble within Ring 2. In fact, on Day 0, it
still is
> _all of_ Ring 2. The name here is mean to evoke Creative Commons; we can
> discuss other possibilities too, of course.
What happens when a RPM (presumably) package from Fedora Commons
depends on infrastructure that decides to move to a stack and stop
using RPM?
The same thing that happens in Fedora now when someone decides to stop
maintaining an RPM.
> In the future, a namespace approach
> could even make /usr/bin/python be Python 3 for some processes but not for
> the core OS.
(Namespaces are an administration nightmare because paths suddenly
don't mean what they seem to mean.)
That's why "in the future". That particular idea is something some people
are exploring but which needs a lot of details worked out. I think we should
encourage the exploration, though.
--
Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ <mattdm(a)fedoraproject.org>