Richard Rauch wrote:

> * Plot a chart of some kind showing different OSs and archetecture
>   types---and measure how long, for each combination, it took
>   for each port of each OS to go from "I'm doing an <x> port" to
>   "It's booting multi-user".  Maybe throw outliers at one or both
>   ends and compute an average time-to-port for each OS.  "If you
>   want a port of a UNIX-like OS to a new platform, here's how long
>   it takes..."
>  
>
Not that many people outside the NetBSD community think portability as a 
priority. I'm sure that chart would benefit NetBSD, but it would seem to 
others as NetBSD-biased. Most new platforms come from the embedded 
market, and most UNIX-like OSes are aimed at the server market. Many 
would choose (and have chosen) features and performance and/or security 
over portability, because of the server market.

> * How many vulnerabilities have been found and over what time
>   frame?  (Pro-actively finding bugs is good.  If the code base is
>   common to other OSes, the bugs should tend to be common as well,
>   if the development teams are comparable.)
>
>   I mention this since, more than once, I've seen someone assert
>   that NetBSD has fewer bugtraq vulnerabilities than any other
>   OS.  I'd like to see that quantified.  (^&
>  
>
IMHO a good comparison would be the time frame of solving 
vulnerabilities comparable with other OSes.
For example, when an Apache vulnerability is discovered, how much time 
it took to be solved?

Number of reported NetBSD-specific bugs are not a good comparison 
method, since fewer reported bugs could be seen as smaller userbase, as 
an hobby or research OS instead of a real-world OS.

>Any current-state comparisons to other operating systems will be
>prone to being inverted.  But recurring things (like twice being
>chosen for setting the networking speed record) could be good,
>as well as longer-term statistical observations.
>  
>
Windows as been choosen for speed records as well. Honestly I think 
things like how much time it took to support new hardware interfaces, 
like USB and others, are better as advocacy. Also network speed has 
allways been part of FreeBSD advocacy with their TCP/IP stacks.

pkgsrc, RCng, portability, and also security and performance - I hear 
lots of times that all NetBSD needs is an NVIDIA 3D driver :D

But most of all I think we need to get all these features into the hands 
of users, convince them they can handle NetBSD: this can mostly be done 
with documentation that helps new users set up their own desktop. Once a 
desktop is set and users feel more confortable with it they'll start 
exploring more and enjoying the benefits of consistent code.


Bruno