Subject: Re: What is 'scripting' in sysinst for?
To: None <tech-install@netbsd.org>
From: David Laight <david@l8s.co.uk>
List: tech-install
Date: 10/07/2003 08:32:37
> 
> What we really need is a nice install framework that does the install via
> some form of configuration specification (done this in the past). 

But I'm not sure having sysinst try to write a shell script is the
way to start.  Adding the main commands to the log file - so they can
be extracted into a script in one thing (at least they won't be expected
to be absolutelt complete - you might log the output of fdisk and disklabel
instead on the commands required to make the changes).

If you are cloning systems, then you need some of the stuff the sysinst
'leaves to the user' done in the clones.

I started this thread after looking at the code that saves /usr/X11R6/bin/X
during upgrade, thinking it would be better to use readlink() and
symlink() instead.  In other places unlink() could be used instead of
system("rm").  However these sor of changes break the script.


	David

-- 
David Laight: david@l8s.co.uk