Subject: Re: HEADS UP: migration to fully dynamic linked "base" system
To: None <email@example.com>
From: Andrew Cagney <firstname.lastname@example.org>
Date: 08/28/2002 10:59:35
> On Tue, 27 Aug 2002, Andrew Cagney wrote:
>> As for /recover:
>> - it isn't used by default so how does anyone even know it works? At
>> least a normal boot gets to work out the normal binaries and boot paths.
If a static executable is never run, can the kernel load it? :-)
>> - remembering to type /recover/init or /recover/sh is going to, er,
>> confuse people and feel very strange to (umm) unix bigots.
Yes, having it suggest /recover/sh or /recover/init would also help.
> We will have a bit of educating to do. I think AB's idea of noticing if we
> succeeded wiht the fork is a good one, and we might auto-fallback to the
> rescue tools easily
>> - does ``init'' really need to be internationalized?
> Maybe, maybe not. But init CERTAINLY can benefit from nsswitch modules and
> auth modules.
Does that block of code need to be part of init? If authenticaion is
needed, run a separate [shlib] authentication program.
> There are three things which programs are benefiting from: easier locale
> support (Asian languages are the ones that make the most difference),
> nsswitch modules, and eventually loadable auth modules. Different programs
> benefit from different of those advantages.
I believe that this is has been accepted as a good thing. The concerns
apear to be focused on the boot sequence (in my mind, everthing that
goes into getting /usr (/var) mounted) and ensuring that it is robust.