| 1 |
Brandon Hale <tseng@g.o> said: |
| 2 |
|
| 3 |
> On (01/20/04 14:47), foser wrote: |
| 4 |
> > |
| 5 |
> > Tseng sees the real desktop issues as secondary, while the installer is not |
| 6 |
> > a desktop thing as such (you never see it beyond your install). I |
| 7 |
> > personally think such a vast project (it is a lot) is really beyond the |
| 8 |
> > scope of this team and at least is not a good way to start defining what |
| 9 |
> > this team is supposed to do. You yourself imply that any installer release |
| 10 |
> > up to at least 2004.1 is very unlikely, I'm afraid it's gonna suck |
| 11 |
> > resources from places where the desktop needs them. |
| 12 |
> |
| 13 |
> Installalling a desktop is a major part of the use experience between |
| 14 |
> distributions. Having a GUI installer is what I see to the the most requested |
| 15 |
> feature from our users, who imo should have a large drive in our development. |
| 16 |
> Also, I simply asked for desktop research to discuss this topic at the |
| 17 |
> meeting, they chose it as a topic for further review without me present. I |
| 18 |
> asked it to be clear that I was not aiming for the actual coding of the |
| 19 |
> installer as an immediate atainable goal, this has happened and failed |
| 20 |
> several times already. |
| 21 |
|
| 22 |
> |
| 23 |
> What I asked is for this excellent research team to draw up clear |
| 24 |
> expectations for the installer, what we want it to do, and create a roadmap |
| 25 |
> for realistic completion. This will allow us to find the skilled resources |
| 26 |
> needed to reach milestones, rather than isolated developers w/ their own |
| 27 |
> incompatible visions of the installer. |
| 28 |
|
| 29 |
I think that's the way to go. Some brain storming (some of that has already be |
| 30 |
discussed here): |
| 31 |
|
| 32 |
personnal idea : have a minimalistic installation (it'll configure only what it |
| 33 |
needs to be installed). The rest is left to the gentoo config tools after the |
| 34 |
installation. If needed, the first boot can be special, with easy |
| 35 |
access/presentation to the config tools. Major argument : don't develop the |
| 36 |
configuration tools 2 times (one in install, one after install). |
| 37 |
|
| 38 |
So what's left : |
| 39 |
configure hard drive, partitioning, network, choose minimal installation to do |
| 40 |
(masked unmasked, etc), root passwd, user additions. |
| 41 |
|
| 42 |
You end up with a minimal gentoo installed, ready to be configured, and |
| 43 |
installed. X might be emerged by the installer, but left in default |
| 44 |
configuration that works almost everywhere. On the first boot, X shows up, and |
| 45 |
propose you to condifure and install stuff (among them kde, gnome). It can be a |
| 46 |
wizard. No window manager needed. You should have the same kind of things if |
| 47 |
you choosed not to have X emerged by the installer. |
| 48 |
|
| 49 |
The goal : do the minimum at install time, -> less work. |
| 50 |
|
| 51 |
hardware detection : you only need to detect : harddrive, cdrom, network card, |
| 52 |
modem, |
| 53 |
usb modem, usb storage/harddrive, floppy, very basic video card detection, usb |
| 54 |
cdrom, usb adsl modems, mouse, keyboard, and thatt's all ! |
| 55 |
|
| 56 |
The beauty of that, is that because you need to detect very few things, you can |
| 57 |
do that well. And add unusual detection, like usb cdrom, network boot and so on. |
| 58 |
|
| 59 |
-- |
| 60 |
dams |
| 61 |
|
| 62 |
-- |
| 63 |
gentoo-desktop-research@g.o mailing list |
Archives