[nSLUG] ubuntu iso a dud

George N. White III gnwiii at gmail.com
Sat Feb 24 10:24:23 AST 2007

On 2/23/07, Robert Ashley <rb.ashley at gmail.com> wrote:

> Appreciate the follow-up. I seem to be doing fine with Debian Etch.
> Or, it's just putting up with me.
> So like I remember a couple of years back when first exposed to Linux
> and this group. Most of the sophisticates advised sticking with one
> distro over flitting from one to the other. So I did.
> That said, maybe I've gone the other extreme--Debian-welded. I've
> never even tried anything else.  A real shut-in, I guess.

The important thing is that you aren't missing out on anything
important by sticking with debian.  Red Hat and other RPM-based
distros are suffering from shortcomings of the RPM package manager.  A
new project to fix RPM has begun, but the transition is likely to be
painful.  As a user you don't need unnecessary annoyances from a
broken PM.
At work we have RH because NASA uses it for a mission critical app,
but I have debian etch on an old PIII system that is used for things
like keeping tabs on UPS's and a RAID via serial cables.   It
installed easily and just works.

At work I'm forced to deal with Windows, and at home my wife uses
Apple OS X.  The good thing about Windows is that there are so many
people using it that someone will have dealt with any problem you
find.  The bad thing about Windows is that there are so many problems.
 There are some really important tools: ghostscript, perl, ruby, TeX
(maths typesetting) that have huge numbers of users running Windows,
but the key developers don't do Windows.   With TeX, there have been
many ports and commercial implementations for DOS and all versions of
Windows.  Very few of these systems have remained viable over the long
run.  The main issues have been:

1.  the open source community has been unwilling to maintain a common
source for both *X and Windows, so each new *X version has to be
ported to Windows.  Some people think that porting tools to Windows
hurts open source, but I tend to feel that when we get to point that
people are using open source tools on Windows they will see the
advantages of moving to *X.

One of the problems for porting *X apps to Windows is that there are
not consistent policies for things like $HOME.   Individuals are
likely to have HOME=C:\Documents and Settings\joe_user, but large
sites and student labs may want user data on a USB key or network
drive so
any user will get the same environment on any machine, so then
HOME=U:\ and the login scripts copy information to the registry.
Unless you know the local policies it is impossible to know where to
put data so it will be available across sessions/systems.

*X policies evolved over a long period.  The people who developed
Windows could have learned from unix and VMS, but didn't.  People have
found ways to port GNU apps (see GNUWin32) so these days the
portability problems are related to package management and mapping *X
policies onto Windows.

2.  massive numbers of users require massive support resources.
Having contributed to the pdfwrite device in ghostscript, I know that
requests for help from Windows users vastly exceed the requests from
*X users and that many of the problems have nothing to do with
ghostscript (e.g., all batch scripts fail or take hours to load due to
a 2Mbyte PATH variable).  It has taken huge amounts of time, but the
result of the open source community supporting pdf tools for Windows
has been that Adobe no longer "ownz" PDF.  Open souce tools are
sufficiently important elements of a PDF workflow that Adobe can't
make unilateral changes without damaging the PDF/prepress ecosystem
that supports a major segment of their business

On *X you can write an app, e.g.,  to install additional TeX fonts
from the network for an individual user.  The system tools will be
able to find the fonts.  On Windows the TeX system would have to find
ways to locate the user's persistent data area before this would work.

There are lessons here for people who want to see Linux become mainstream:

1.  viligence is the price of freedom -- it takes work to ensure that
sources remain portable and standards compliant.

2.  success means more users and particularly more neophyte users who
require support.  For open source to become mainstream,
systems have to be much more robust or support demands will be overwhelming.

3.  supporting legacy hardware is a huge opportunity as well as a
liability.  Each new version of Windows needs a massive increment in
the hardware.  People are discarding P-III systems because it takes
hours to load MS Outlook.   Linux can work nicely on such machines,
but they also have a high rate of hardware failures, so linux support
resources become "after warranty" hardware support for Dell, etc.

George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia


More information about the nSLUG mailing list