Re: status of Progeny projects
On Sat, 2003-11-01 at 09:49, Philip Brown wrote:
> On Wed, Oct 29, 2003 at 10:10:19AM -0600, Gunnar Wolf wrote:
> > Besides, last time I checked, Anaconda was HUGE - Would not
> > work if we want to still provide the ability to install using a couple
> > of floppies.
>
> that's only a problem if you insist on having the exact same install
> programs on the floppy set, as you do on the cdrom set.
>
> These days, I think that is not a good goal to aim for.
> Better to aim for cdrom as the "normal" install, and provide a cut-down
> version of install progs, for floppy install.
>
> Or in other words: migrate to using anaconda (or something just as
> interesting) for cdrom installs on systems that can support it,
> and the current stuff otherwise.
>
It doesn't work that way. Its very hard to cut down a large program
written to take advantage of lots of resources (other programs available
on CD, lots of memory, etc); much more possible to do a scalable design
by building _up_ as you get more resources: a system that you drop in
other pluggable modules as you have the resources for them. This is the
idea behind d-i.
Secondly installation is something that needs lots of testing in lots of
environments: having cut-down special case programs for i486 /
low-memory/ arm / ... is a recipe for them not being tested properly.
Sharing the same (core) modules amongst all will see them better tested
when running in strange small environments.
And such environments are important: the free software world spreads
beyond the PC, into PDAs, etc. that can increasingly be built with
Debian, with minmal changes to d-i for your special project. We should
not assume all hardeware has CD-ROMs when some new laptops don't even
have them - eg d-i allows for USB memory stick installs.
- Alastair McKinstry
Reply to: