To: Dennis D. Sherod <>
From: John F. Woods <>
List: current-users
Date: 01/14/1994 11:14:25
>> I really don't like the idea of putting a dependency on perl into the NetBSD.

Well, even having suggested perl as a temporary solution, it's not the only
one (indeed, I don't use perl myself).  But for the specific complaint (suid
shell scripts were removed and won't be replaced immediately), it was a

>IMHO, decisions like this should take into consideration of who the enemy is... 
>Windows/NT.  With that in mind, I think perl would be a good place to start to 
>make things better, but I don't think it goes far enough.  I'd suggest to all 
>the O/S development groups out there, including NetBSD, that X should be part 
>of the base system along with tcl/tk for administration duties.  It would be 
>nice to  see a second-stage bootstrap initialization that came up in some tk 
>application to get the rest set up on a machine.

(1)  tcl/tk may be awfully sexy, but it's even further away from being a
mainstream small-unix-system tool (I haven't tried it, whereas I have tried
perl :-).
(2)  I would venture to guess that a substantial majority of the NetBSD
developers *use* X, so it is unlikely that NetBSD releases will be incompatible
with contemporary versions of X, but X is a substantial body of software, and
expecting the NetBSD development team to productize X as well as NetBSD is
asking rather a lot.  *None* of the freely-available OSes, maintained and
distributed by hardy volunteers, are going to defeat or even dent the Windows
juggernaut; there are far too many people for whom "grab the binaries by ftp,
and if you have any problems, make sure you have a large enough disk for
sources and a netnews feed for comp.os.386bsd.*" is a vastly inferior strategy
than "buy it at K-mart and if you have any troubles, call this 800 number."
At best, the freely-available OSes can hope to be credible bases for small
companies to provide integrated and supported distributions (supported as in
providing marketing and handholding).  Indeed, BSDI demonstrates the validity
of that approach (which uses Net-2 as their base), as does Cygnus (which sells
primarily support and cutting-edge software changes to GNU software).  

I think the NetBSD team has the right idea; a solid base that hardy computer
experimenters can use directly, and which lends itself to integration work
by third parties for the benefit of people who just want do make their ^&*#%$
computer do something vaguely useful for them :-).  The -current list has
gone over the question of where the dividing line of available-with-NetBSD
and works-with-NetBSD-you-know-where-to-find-it should be; there was general
agreement that X was way, way over that line, because of its size and because
there's already a separate group diligently working on X as a "product"
(i.e. the XFree team).