upwards and onwards

[Matthias Clasen]

On Fri, 2009-11-13 at 16:06 +0000, Colin Walters wrote:
> On Fri, Nov 13, 2009 at 3:24 PM, Matthias Clasen <mclasen at redhat.com> wrote:
> >
> > - Drop the CD size limitation and target a larger usb stick. We have
> > discussed this in the past. The main motivation for this is that we have
> > to fight every release cycle to make things fit on a CD, and we don't
> > have room to include our default office suite or example content.
> 
> So...let me counter-propose this one with a multi-pronged solution:
> 
> * Continue trimming the default install.  Adam did some work on
> removing Perl (where were we on that)?  The base comps group has grown
> all sorts of arbitrary stuff.
* Include a PackageKit "Complete your installation" hook post-install
> which does "yum groupinstall @gnome-desktop" 
> * Elevate the livecd inside Fedora to be more of a primary product; if
> someone uploads a package which takes significantly more space, we ask
> them to revert or fix

This is all fine and cool, but it is not going to help at all around the
basic fact that an ever-growing percentage of the cd (is it 30% ?, maybe
even more) is eaten by localization (translations, fonts, dictionaries).
And we are still not going to be able to include example content, eg. a
movie to try totem.

> And in addition to these, we generate a 2GB sized "full" install
> (hopefully @gnome-desktop fits in 2G...) which could also go on a 3+GB
> USB key or a DVD.  This image is spun as the "Full" installer.

Noo, I don't want to see more alternatives. We already have the big
problem right now that people choose to download the largely
unmaintained DVD, just because it is bigger, and thus must be better...

Also, how exactly do you think a 'full' install is going to differ from
the regular live image ? Mostly, it is going to add redundancy and
things that we didn't include for a reason, no ? 

I don't think we want to have a situation of
Live CD - minimal install, missing important things
Full DVD - all the crap

I'd rather see us work on a single image and try to make that as good as
we can, instead of working on two suboptimal images.