Re: Trying to break the anaconda/distro embrace

[Greg Dickie <greg max-t com>]

Sounds right to me. 

Greg

On Tue, 2006-12-12 at 15:50 -0500, Jeremy Katz wrote:
> For a long, long time, anaconda has been pretty closely tied to a
> specific version of Fedora/RHEL/RHL.  This is something which has
> slowly been getting a little bit better.  As we continue to move
> forward and try to get contributions of other backends[1], I think
> that it's important we continue to move in this direction. 
> 
> As it stands today, if you boot with the rescue CD, you can enter an
> arbitrary repository path and be able to do an installation with the
> packages from there without any other real dependencies.  I think that
> this is something which is very very valuable.  Much more so than the 
> existing boot.iso/diskboot.img. 
> 
> 
> Given this, I'd like to propose some changes for anaconda over the
> timeframe of the next Fedora release:
> 
> 1) Replace boot.iso/diskboot.img with an image analagous to
> rescuecd.iso (installcd.iso)
> 
> 2) Make it so that selection of software repositories can be deferred
> to the second stage.  There's already some basic repository selection
> for extra repositories; this would just be extending that to show the
> "base" repository as well 
> 
> 3) Instead of pointing to an installation tree in the loader (for pxe
> installs and maybe still diskboot.img), you should point at an
> "installer location".  This will definitely include the location of
> stage2.  It may also be metadata to point to an installation tree so
> that we can preserve existing behavior. 
> 
> 4) Split out the majority of what's currently under scripts/ in
> anaconda CVS into their own package/repository.  This should contain
> the distro specific bits related to creating an install image.
> 
> 5) Make sure that the scripts in 4 make it easy to just create
> an installcd.iso.  Then, while we may want to have them generated with
> the daily rawhide push for Fedora, we can also keep versions around
> for people to test with and also make it easier to push an "installer
> test"
> 
> 
> Does this make sense to people and seem like it would help some in
> developing changes as well as doing testing for anaconda?
> 
> Jeremy
> 
> 
> [1] Hopefully we'll start getting some of the ConaryBackend bits in.
> I'm also going to probably be throwing together a backend to install
> from an "image" with live CDs.
> 
> 
> _______________________________________________
> Anaconda-devel-list mailing list
> Anaconda-devel-list redhat com
> https://www.redhat.com/mailman/listinfo/anaconda-devel-list
-- 
Greg Dickie
just a guy
Maximum Throughput