Subject: Re: NetBSD usage in embedded environments
To: None <>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: tech-kern
Date: 04/03/2003 01:30:26
>> I [...] asked, "Why a full blown OS instead of some smaller
>> real-time thing?"
>> He replied, "because a megabyte of RAM only costs $1."
> it makes sense not to put in code that is never going to be used.

Yes.  But it also makes sense not to bother cutting code that is never
going to be used, if the cost of preserving it is lower than the cost
of cutting it.

That is, there's code that's _put_ in, and there's code that's _left_
in.  Going embedded with NetBSD, there are often lots of pieces that
are strong candidates for the latter.

> the folks who port netbsd may not (always) be v. good at gleaning out
> unnecessary code.

Or they may just, quite literally, have better things to do: the cost
of shipping, for example, whiteout support in ls(1) consists of the
disk (or equivalent) space to store the code, plus a little RAM when ls
is run - even if your device will never use whiteouts.  That may be
lower than the cost of paying someone to go in and rip it out every
time you re-import a new NetBSD version.  (It may be higher, too, of
course, depending on various factors.)

> so, a little bit of work in adding some more config options will go a
> long way in removing redundant code from numerous devices.

Perhaps.  What would you suggest?  To return to your first example, I
don't think adding a config option to dike out the swapper is "a little
bit of work".  Maybe I'm wrong - show me. :)

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B