Subject: Re: Something to shoot for someday...
To: Chris G. Demetriou <>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
List: tech-install
Date: 07/01/1999 23:48:22
>I don't want to discourage other people (e.g. you) from trying to go
>the "large rewrite" route, but evolution is, at most times, more
>practical than revolution.  8-)

And the weird thing here is, i've been consistently suggesting
evolving sysinst via a style where most of the backend would have a
much LISPier or Scheme style: lots of little functions to deal with
newfs'ing and untarring and config files and binary formats (like
disklabels).  with wrappers around them to handle scripting, and
logging, usw. 

Last: my impression is that most of the actual problems with sysinst
are due to the x86 MBR handling.  Lest some people forget,
most of the feedback on 1.3/pmax was along the lines of:

    ``Hi, I tried instlaling NetBSD on my pmax. Got through sysinst.
      No sweat!  But then when I go multi-user I'm hitting a race
      condition with the Xserver and xfs. How do I fix that?''

so sysinst, per se, cant be quite as inherently catastrophic as some
people make out.  True, it doesnt look like those Lizardware
screenshots, but nothing curses-based ever will.

Or are we deciding to  abandon all platforms that dont have SVGA?