The Trolling on the freebsd- lists

Gary Thorpe gathorpe79 at yahoo.com
Tue Jan 6 12:03:36 PST 2004


Bill Huey (hui) wrote:

On Tue, Jan 06, 2004 at 11:12:22AM +0100, Miguel Mendez wrote:

Also note that discussing FreeBSD's merits in this list is off-topic. 
The last time DragonFly was discussed in the freebsd-hackers@ mailing 
list, Kip and other people got flamed by Poul-Henning and friends.

...which is precisely the problem of the FreeBSD community. Instead
of discourage forking, it should encourage it and then be supported
under the same umbrella as if it was the same project with no intrusive
centralization.

Functional groups, with many highly qualified engineers, ideally should
let things be a plurality without problems or find ways where as many
ideas can be expressed without exclusion. Obviously, folks like phk and
others prefer to put down other potentially interesting work to preserve
their sense of CVS territory and artificially boost their own status
instead, pissing off the very people they need for the project to be
successful.


I would like to point out that branching is permitted and encouraged 
under Linux, which although has been around for less time is currently 
more advanced in many (not all) areas than any of the BSD's. 
Unfortunately, there are also examples of concensus/politics taking 
precedence in Linux also such as NGPT vs NPT thread model, in my opinion.

I would think that since CVS was designed with this in mind (branches, 
alternate threads of parallel development) that it would be a natural 
outcome. Since work on branches will never directly affect each other, I 
don't see why the main branch could not hold the 'important' code and 
yet keep experimental branches around to try new ideas. As far as I 
know, this sort of thing only happens normally for the Linux kernel 
(which paradoxically, did not use version management in any form until 
the last few years): none of the BSD's do this (yet). Am I mistaken in 
this general observation? Why has it happened this way for Linux/BSD? 
Note: having a current branch (the head) and stable branches for 
releases is not what I am talking about. For example there where 
alternate, competing VM designs in Linux in the 2.4 series (different 
branches) as well as the Alan Cox (ac) kernels.

I keep wanting to exit this thread, but I feel this needs to be said
repeatedly somehow.
bill


Is it just a problem with FreeBSD? Could there be concerns that 
developer time/effort will become too divided and if so are they valid?








More information about the Kernel mailing list