*BSD News Article 20933


Return to BSD News archive

Xref: sserve comp.os.386bsd.questions:5086 comp.os.386bsd.misc:997
Newsgroups: comp.os.386bsd.questions,comp.os.386bsd.misc
Path: sserve!newshost.anu.edu.au!munnari.oz.au!constellation!osuunx.ucc.okstate.edu!moe.ksu.ksu.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!noc.near.net!lynx!random.ccs.northeastern.edu!news
From: jtsilla@damon.ccs.northeastern.edu (James Tsillas)
Subject: Re: A merge of FreeBSD and NetBSD? (Another person's opinion)
In-Reply-To: haley@husc8.harvard.edu's message of 12 Sep 93 18: 31:42 GMT
Message-ID: <JTSILLA.93Sep12213741@damon.ccs.northeastern.edu>
Sender: news@random.ccs.northeastern.edu
Nntp-Posting-Host: damon.ccs.northeastern.edu
Organization: Northeastern University, Boston, MA
References: <1993Sep8.231610.9740@ccds3.ntu.edu.tw>
	<CD190K.FwG@latcs1.lat.oz.au><CD3JII.F5w.1@cs.cmu.edu>
	<26p8ul$1eb@pdq.coe.montana.edu>
	<MYCROFT.93Sep11213749@duality.gnu.ai.mit.edu> <haley.747858702@husc8>
Date: Mon, 13 Sep 1993 02:37:41 GMT
Lines: 50


Having two very similar (if not three) operating systems seems to me
like a incredible waste of personal resource and will give rise to much
duplicate (triplicate?) work for what appear to be simply religious
reasons.

There are no good reasons that I can see for keeping the efforts
separate. The arguments that we need an experimental release of the OS
in parallel to the stable release is no excuse for the parties to work
independently. What is needed is a good source control/configuration
management system and some good will among all the parties involved. The
first is easy to obtain, the second is much more difficult.

Professionally, I work as the SC/CM specialist amongst a group of thirty
software developers. Our product is the network OS of a high speed
router-bridge. It is not uncommon for us to have three major release and
one beta release in the field and under development (whether
maintenance or new development) at any given time. In addition to this
we have some five or six divergent/experimental efforts towards the
development of new features which our customers demand. We are also
pursuing different hardware architectures and there is usually at least
one porting effort ongoing. Our customers demand this aggressive release
and development policy and we've put alot of resources into perfecting
the underlying methodology and process. My point is: it can be done. You
can have the experimental branch and the stable branch working side by
side and borrowing from each other in a coherent and managed way. It's
not easy, and it takes at least a few dedicated individuals to keep
things on track. But the results are much, much better than keeping the
efforts separate.

I urge the developers of all 386BSD based efforts to carefully consider
the benefits of a parallel release strategy using a good SC/CM system. I
urge them to give a high priority to merging the two OS's and to reach a
common code base. From there the source can be entered into a SC system
to provide a branch structure for the parallel efforts. The major (or
"main" branch) can be used to keep the latest stable source for the next
stable release while the several minor branches can be used to keep
unstable source for experimental releases which may be merged to the
major branch on a selective basis. I believe that RCS can provide 100%
of what you need to make this happen.

regards,
-Jim.


--
	***  James Tsillas  jtsilla@damon.ccs.northeastern.edu   ***
	***      Work: (508)898-2800, Home: (617)641-0513        ***
	***        "He is after me. Jim is after him."           ***
	***            - Hop on Pop, Dr. Seuss                   ***