On Wednesday 03 December 2008, Frans Pop wrote:
> multi-arch images
> -----------------
> The i386/amd64/powerpc DVD is more problematic as having all desktops
> will completely fill the DVD leaving no room for other packages with a
> high popcon score.
> Proposal is therefore to drop powerpc support from this DVD. With only
> i386 and amd64 there is still room for the top ~2600 packages on the
> DVD.
I made a mistake here as I forgot that multi-arch (m-a) DVD images are
also supposed to include source. So let me give the correct data and then
draw some new conclusions for this image.
First, let's take a look at what's roughly on existing images.
The 3rd column is the existing i386/amd64/ppc/source DVD. The 4th column
is a fictional i386/amd64/source DVD, i.e. what would happen if ppc were
dropped from the current m-a DVD.
amd64 CD1 ppc CD1 m-a DVD (+ppc) m-a DVD (-ppc)
* key packages
- GNOME all all all all
- Server tasks all a few all all
- l10n all none all all
- l10n desktop all none all all
- l10n GNOME all none all all
* non-key packages
- GNOME a few none most all
- desktop (OOo etc) none none none all
- Server tasks none none none all
- l10n none none none all
- l10n desktop none none none a few
- l10n GNOME none none none none
- l10n KDE - - - -
Note that there is a HUGE difference between the amd64 and powerpc first
CD images: the second is incomplete to the point of being unusable by
itself. Reason is the number of different flavors included for powerpc,
which results in multiple D-I initrds and kernels and multiple regular
kernel packages being included.
The same effect can be seen in the size of the netinst images:
- i386: 150 MB
- amd64: 131 MB
- powerpc: 198 MB - 33% larger than i386!
Note also that the current M-A DVD (including ppc) almost exactly matches
the amd64 CD1, which is roughly what was intended. But it is also clear
that dropping ppc from it would increase its usability for installs
without network mirror by a huge margin.
So now let's look at how the m-a DVD would look if we'd include all
desktop environments. The first column lists the existing m-a DVD for
comparison.
GNOME desktop all desktops all desktops
m-a DVD (+ppc) m-a DVD (+ppc) m-a DVD (-ppc)
* key packages
- GNOME all all all
- KDE - all all
- Xfce/LXDE - all all
- Server tasks all all all
- l10n all all all
- l10n desktop all all all
- l10n GNOME all all all
- l10n KDE - all all
* non-key packages
- GNOME most a few all
- KDE - none all
- Xfce/LXDE - none all
- desktop (OOo etc) none none some
- Server tasks none none none
- l10n none none none
- l10n desktop none none none
- l10n GNOME none none none
- l10n KDE - none none
So an all-desktops m-a DVD without powerpc does a lot better job of
actually supporting all desktops than if powerpc is included; l10n
support is limited in both cases.
I think we have the following options here:
1) keep the m-a DVD as it is: GNOME only and i386/amd64/ppc/source
(3rd column in 1st table, or 1st column in 2nd table)
2) drop ppc from existing m-a DVD, GNOME only, but with hugely increased
usability for installs without network (4th column in 1st table)
3) change m-a DVD to support all desktop environments; drop powerpc
(3rd column in 2nd table)
4) something I have not investigated: drop source...
My personal vote goes to option 3.
Cheers,
FJP
Attachment:
signature.asc
Description: This is a digitally signed message part.