[nycbug-talk] Meeting Feeler: Non-BSD projects Using BSD software

Matthew Story matthewstory at gmail.com
Mon Mar 12 18:54:00 EDT 2012


On Mon, Mar 12, 2012 at 7:41 PM, Pete Wright <pete at nomadlogic.org> wrote:

> On Mon, Mar 12, 2012 at 12:21:10PM -0400, Matthew Story wrote:
> > On Mon, Mar 12, 2012 at 11:30 AM, Edward Capriolo <edlinuxguru at gmail.com
> >wrote:
> [...snip]
> how about closed/propritary systems which use BSD code.  I think that
> is one of the greatest benefits of the BSD code in the real world.
> While things like GNU/kFreeBSD are interesting to look at - the list of
> very high profile systems running real production traffic that
> incorporate BSD is code is both long and impressive:
>
> (off top of my head)
> - NetApp Data OnTap
> - Citrix netscalers
> - bluecoat systems
> - juniper systems
>
> this does not even touch companies that run modified BSD code in their
> own systems with out advertising it.
>

This is an interesting angle that I hadn't considered.  Many (if not all)
of the above are actively devoting resources to the open portions of the
systems as well, but I would be interested in hearing more about how they
are modifying BSD code, and what the benefits are of working from that base
... if they are willing to share, of course.


>
> > While I will continue to run FreeBSD personally, and very much like the
> > fact that it is a complete OS, with a coherent and non-political
> objective
> > ... as a system, I think that *BSD projects could benefit from
> encouraging
> > those who are interested in running components of this system (either
> > kernel or userland) married with other components.  Conversely, I think
> > those involved with other projects could benefit greatly by marrying
> > portions of *BSD systems into their own systems.
> >
>
> i personally am uneasy with projects like GNU/kFreeBSD and the like.
> one of the other huge benefits of a BSD unix is that it *is* a complete
> OS, not a kernel with bolted on userland.  as such i've never understood
> the utility of GNU/kFreeBSD.  i'm probably biased, but after working on
> a Nexenta system (which seems pretty similar to gnu/kfbsd) i ran crying
> as fast as i could and just put solaris back on the hardware.  it always
> seemed like so much was lost just so someone could say that bash was the
> default shell, and look it has apt.
>

I agree that BSD systems are better largely because they are ... well ...
systems.  But I think it's a mistake to not advocate making use of parts of
that system in other systems.  My reasons for this are 2-fold:

1. More people using BSD software means more eyes (and hopefully more
hands) looking (and hopefully developing) on code, and theoretically this
means better code long-term, faster iteration on bugs, and generally better
maintained software.  This point boils down to it being better for BSD.
2. BSD software is simpler than other software, and most of the time
better.  This point boils down to it being better for other projects.

There may be some issues with tunnel-vision resulting from the former, e.g.
someone is only interested in NetBSD shell, or FreeBSD SMPng, and not
interested in the overall health of NetBSD and FreeBSD as systems ... but
this seems to me to be a small problem compared to the benefit of more eyes
and hands on code.

One way to improve the strength of the projects is to advocate the use of
the system, and I certainly try to do this.  But if someone is married to
the Linux Kernel for legitimate reasons, or to GNU user-land for legitimate
reasons, why not invite them to make use of parts of the system when they
cannot use it whole hog?

-- 
regards,
matt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nycbug.org/pipermail/talk/attachments/20120312/ca26763a/attachment.html>


More information about the talk mailing list