Subject: RE: Why so many BSDs?
To: Dave Barr <firstname.lastname@example.org>
From: Christian Gruber <Christian.Edward.Gruber@gmx.net>
Date: 12/01/1999 12:06:30
And how exactly does name-calling, in the name of moderation, benefit the
cause of Free-UN*X unity?
One of my big problems in the wars between OS's is not that facts are
distorted, as those facts can be countered by better facts, but rather that
the rhetoric and religion flows thick, preventing anyone from even getting
to factual arguements.
Theo's being a nut or not is irrelevant to the facts you mention. Secondly,
Theo was paraphrased without references, so holding him to the statement is
like hearing someone say, "I heard that Clinton (Dole... whoever) hates
black people," and then replying "What a bastard" instead of "Quote your
source." It's bad science, so to speak.
Mostly this personal thing really pisses me off, because even if Theo were
as nutty as you ascribe of him, you contribute to extremism in the issue by
calling him so.
A few points:
> 1. SuSE, RedHat, Debian, Slackware are the big 4.
> They account for the vast, vast majority of the
> installed base.
1. Windows has most of the install base, so the
arguement of 80/20 may make some sense, but
has a real problem when taken too far.
> 2. Many so called "separete distributions" are
> either total image copies of the above or the
> above plus some amount of branding or additional
> software packages. To say they are a "separate
> distribution" to the same degree this has
> meaning as say, RedHat versus Debian, is a lie.
2. Agreed, but you must remember that if I use one
distribution, which is red-hat-like, and they change
at v.X to use a slackware style packaging system,
I have a choice. I can change distributions blah
However, the fact that they can alter their
distribution without synchronizing modifications,
means that they're different enough. From a
Project-management and Product-management standpoint,
these issues of revision control show up as a severe
quality assurance hassle. Linux has the wonderful
advantage of having people who will rip-off any
good changes from such an altering distrubution,
and merge them into the whole, and will probably
ditch any distribution en-masse if that org makes
mods which render their version incompatible... but
to use the software in a commercial or other setting
where stability and security are required, some
level of stability needs to be guarranteed. Now
paying red-hat some money will probably guarrantee
it, and that's one legitimate way... pay for the
service, and Red-Hat's reputation requires them to
be careful. But that's a different dynamic than
an actual approach which encourages stable merging.
Which is a better system, BSD or GNU? Who the heck
knows, but in one you get fewer branches, and
stability from the process, and the other, you get
more branches, and stability from commercial demand.
Take yer pick.
> 3. Given the same criteria that Theo no doubt uses
> to get the 240 or 260 number, one cannot say that
> there is only 3 versions of BSD. You'd have to
> get a number in the 60-100 range. (add all the
> *BSD flavors, multiply all the branches they
> have, multiply all the architectures they support,
> add all the non-integrated driver development,
> etc). It's easy to inflate the numbers using
> meaningless criteria.
3. Agreed, such inflations are unreasonable, if they
include architectures, and "branches within a project",
but there are fewer independant groups working on
OpenSource BSD-based operating systems. I know of but
a few, including OpenBSD, FreeBSD, NetBSD, and Darwin.
There are some specialized OS's for real-time computing,
massively parallel stuff, but these bill themselves
quite differently, and I think the number is quite
small. I think that the arguement about numbers of
distributions is entirely a red-herring. It's more,
if anything, about process control, and too often,
Well, Theo is a nut, in more ways than one.
You can inflate the numbers to suit your own agenda, as
apparently Theo is (and others are) doing. [snip]