Re: Install report (2003-04-30)
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Joey Hess wrote:
> Petter Reinholdtsen wrote:
> >
> > [Joe Nahmias]
> > > Why throw away that information and start from scratch trying to
> > > detect the hardware, when we know what works?
> >
> > Well, I would turn the question around, and ask why should we hardcode
> > which modules to load when we can load the needed modules dynamically.
> > This way the installation will keep working even if some HW is
> > replaced.
>
> I hope you're not proposing defaulting debian to some kind of kudzu-like
> auto-detect-hw-on-boot thing. If so, I have about 5 acres of urls to
> stories of that failing to dig up and post here.
Agreed, though I don't know how much I count for as a non-DD... :-(
> It's hard enough to get the right modules worked out once, by detection
> (which sometimes fails) and manual tweaking. Trying to get it right
> every boot is not suitable for production machines.
You've made my point exactly, although in a less visceral manner than I
would have :-) This is why I was advocating storing the info that d-i
already has, instead of trying to re-invent the wheel (or rediscover the
modules) every boot. After all, if a user _knowingly_ changes the
hardware in a machine between install and the reboot for base-config, I
don't think it's too much of a burden for them to re-run discover
manually (or whatever) to reflect that change; as they would have to in
a non-installer scenario (principle of least suprise would seem to
apply, no?)
Petter, does this logic convince you? If so, would you revert the
change you made to hw-detect? If not, why? ;-)
Thanks in advance,
Joe Nahmias, (still a) DD wannabe
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
iD8DBQE+suBGKl23+OYWEqURAgvLAKDY5LJUCSNXxivKlMVe/Lo5yJ/0oQCfQtrY
fkAHZSGj0SARlF+cILARhks=
=xhgN
-----END PGP SIGNATURE-----
Reply to: