[Fedora-livecd-list] pilgrim livecd work
davidz at redhat.com
Sun Sep 24 00:19:21 UTC 2006
On Sat, 2006-09-23 at 18:40 -0500, Jasper Hartline wrote:
> > I see. Sorry, but I think this is a little backwards, I mean, why
> > wouldn't I just download the install CD's; it's pretty much the same
> > amount of traffic.
> I'll leave you to debate the difference between a router style 550MB
> install and a full blown everything
> chosen 7GB install. I'm sure a calculator isn't necessary to see the
I'm not sure what point you are trying to make here.
> > Also keep in mind that many consumers of live cd's (typically in poor
> > countries) have zero network connectivity - all they have is a live cd
> > that some benevolent organization sent them. Like Ubuntu's ship it.
> > The approach I have in mind includes just reusing the ext3 file system
> > we have already prepared and just dd that onto the target installation
> > block device. That way an install would take very little time and the
> > only thing you need to make this work is the live cd itself.
> This is extremely unflexible and the user is limited to what you've
> mastered onto the LiveCD
Excuse me? How is this different from Kadishchi built live CD's? CD
space is limited.
(Oh, sure, it's different insofar that you don't support installing
software while the live cd is running - and, you know, this may be the
crucial point of people deciding to install Fedora - that they can try
and see if their favorite piece of software is actually available before
installing Fedora onto their hard disk.)
> without yum installing xyz all day long, which they don't even know exists.
Pirut is included on my live CD and will, as such, be included on the
hard disk when doing OS installs.
> With Anaconda it is a familiar, flexible, choosable option, package
> selectable interface
> along with that upgrading the system from Core to Core is a snap, using
Oh, I see, it sounds like you're making the point that the Anaconda
package selector is better than Pirut. If so the Pirut authors better
I'm not really sure who you are designing for here and it sounds like
you are not sure yourself either. I think that important goals of a live
CD should be
1. Installable without requiring a network
(since many potential users will have little or no network
connectivity, cf. distributions to poor countries.)
2. Ability to install / try out software before installing the livecd
to their hard disk.
(to let people see if the apps they need are available)
3. Contains all software the user base you are trying to target wants
(why it's good to easily be able to build different flavors, e.g.
GNOME desktop, KDE desktop, Eclipse, Music, Games, whatever)
all this while fitting on a single 700MB CD. Besides point 1. I've got
this covered in pilgrim and in a few weeks I should have command line
utilities at least to do the live installs. The UI will follow once the
bespoke gnome-diskutil is ready.
Jasper, why you (obviously) disagree that these are important for Live
CD infrastructure is quite frankly hard to grasp especially when we have
competitors (Ubuntu) that do this already and is kicking our ass in this
department. I also think it's a really bad idea to try to fix this by
using technology (Kadischi) that is inferior to theirs.
Have a nice day.
More information about the livecd