From: Brian Galbraith (brian.galbraith at domain bigfoot.com)
Date: Wed 08 Sep 1999 - 00:48:53 IST
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
This was posted to the SuSe list....I have passed it on as I think it makes
very interesting reading
Brian
- ---------- Forwarded Message ----------
On 06-Sep-99 I (Ted Harding) wrote:
> See
>
> http://www.independent.co.uk/net/990906ne/P18S3.html
>
> Interesting ...
>
> Ted.
>
> --------------------------------------------------------------------
> E-Mail: (Ted Harding) <Ted.Harding at domain nessie.mcc.ac.uk>
> Date: 06-Sep-99 Time: 22:05:52
> ------------------------------ XFMail ------------------------------
It's also interesting to see from the responses so far that Linux users
are not reading between the lines of the article, for the most part.
When I described it as "interesting ..." I didn't mean that I agreed with
it. However, I found it provoked a number of points to ponder.
The article did take the trouble to pay Linux a number of graceful
compliments regarding its performance, etc. I dare say none of us
seriously dispute that.
The main theme of the article was, however, the possible bewilderment of
IT managers in the face of the variety of options, their possible
reluctance to commit resources to one version when it may later turn out
they should have installed another, and also the issue of committing
resources to staff training etc for different GUIs and so on.
I do suspect that this point has some validity. (And it's not just the
GUIs and the icings on the cakes that differ from distribution to
distribution).
We Linux enthusiasts happily (on the whole) accept that using Linux
involves a perceptible element of simply getting it to work (and that
is part of the fun). Spending a lot of effort on "simply getting it to
work" may not be an appealing prospect for a corporate IT manager,
however.
A potentially more weighty factor involves choice between distributions.
For instance, for various reasons I have been through a total of 8
different Linux distributions: in some cases the change resulted from
the necessity of bringing the system up to date (e.g. a.out->ELF),
in other cases from the need for further functionality not supported
in the current installed version. In each case, such a major change
can only be made safely by installing from scratch. Even our home
machines are very heavily used for work purposes, so there is a very
time-consuming effort required to make sure that all working files
custom scripts, special software etc., are saved and ported across and,
where necessary, modified, or else re-installed, so that they continue to
do what they used to do. (It could be seen as an unfortunate coincidence
that the general change from libc5 to glibc=libc6 occurred over about
the same period as coporate IT began to get interested in Linux).
Certainly, when the time comes for me to make a major change of this kind,
I tend to try clean-slate test installations of the 2 or 3 currently
most stable-looking distributions and see which one I get on with best
(or at all -- I recall a recent total failure to install SuSE-6.1
by NFS on a laptop, while both contemporaneous Red Hat and Debian
went on without a hitch by the same route, as had SuSE-5.2 previously).
I can see that an IT manager may in principle be willing to take a step
in the dark, but not relish the prospect that he has to shut down the
shop while he makes a new installation, if it should turn out that he
made the wrong choice in the first place.
So an element of "with Linux, you never quite know what you are getting"
may be understandable, along with an element of "at least with Windows
you know where you stand" (even if, on close and expert analysis, both of
these views may be incorrect).
If Eva Pascoe's article has a valid point, it is a point about
_perception_ and not about philosophical truth. I'm last but one to none
in estimation of her record of technical accuracy, but she may have a
better view of people's perceptions than we may have.
Now to some points which are all mine own. It is my impression that the
complexity and richness of Linux is fast leaving the user documentation
behind. I have 7-plus years experience of installing and using Linux;
yet I find myself poking around in all sorts of documentation corners
even more than I used to, for certain purposes. There are a lot of things
which now install and work smoothly which used to be problems; but there
are also many things which used to be straightforward which now require
research, sometimes considerable research. OK, the mailing lists are a
great help; but it's pot luck.
And where does the Linux Standard Base stand these days? See:
As far as I can tell, exactly where it did six months ago, i.e. almost
nowhere (which was its situation when I looked at it back in February;
of course I could be wrong about its current state -- unfortunately
every single link from the above URL fails to work: I simply get
"connection refused").
A propos of which, I shall close by quoting from a brief (again
unfortunately -- WHY does nobody take this seriously?) exchange of mail on
the SuSE list.
MAIL No 1:
=================================================================
From: (Ted Harding) <Ted.Harding at domain nessie.mcc.ac.uk>
To: suse-linux-e at domain suse.de
Date: Mon, 01 Feb 1999 15:41:50 -0000 (GMT)
Subject: Re: [SuSE Linux] SuSE derailed? Not if...
On 30-Jan-99 Kaare Rasmussen wrote:
>
>> This is what the Linux Standard Base (LSB) project is aiming to, and
>> we are working on this together with the other distributors. It just
>> takes some more time, it's a huge project. Have a look at
>> http://www.linuxbase.org/ for further info.
>
> LSB is a nice effort. It may or may not succeed. What I'm worrying
> about is that for the average person, Linux == RedHat and more so if it
> comes preinstalled. Who will you sell an upgrade to if the people only
> knows another distro?
I hope LSB (or equivalent) will be achieved. I think it is essential at
this stage of Linux evolution for there to be a standard which everyone
can expect, no matter what distribution they choose to install.
At present "Linux", by definition, simply means a Linux kernel plus
whatever else you decide to bundle with it. Contrast it with FreeBSD,
where this is defined by what is one the FreeBSD ftp site.
What is required is a definitive file-system structure (including config
files), a definitive configuration procedure, a definitive installation
and upgrading mechanism, etc, and an absolutely standard linux "base"
system. This should apply to different distributions (though any
distribution is of course free to set itself aside but then must accept
the consequences) and, as far as possible, should apply to any one
distribution consistently over time.
Now that Linux is being adopted by enterprises and institutions, and other
setups whose primary interest is in getting work done, such people will
want assurance that their choice of distribution will continue to be
upgradeable and usable. They do not want to fear that their distribution
may go out of date or have to be changed for another distribution.
They do not want to lose days of productivity time while they move from
distribution to incompatible distribution, or upgrade from version to
incompatible version.
At present, you cannot expect to safely install a new distribution with a
different distribution in place. Even though Red Hat and SuSE both use
RPM and are in many ways similar, there are enough basic differences for
this to be very risky. The only safe way to move from one to the other is
to back up everything you want to keep, and install the new one from
scratch. Still less could you expect to safely install Debian or
Slackware over an existing RH or SuSE (or vice versa).
I remember (with reasons) the sequence of distributions I have been
through since 1992. First, SLS 0.99 (good for its time, but it
disappeared). Then MCC-Interim 1.0: good; but limited: so I moved to
Slackware-1.0 (also good), but this became obsolete (a.out->ELF) after a
couple of years, by which time Slackware was almost out of sight (Patrick
Volkerding had quit). Debian has not quite reached 1.0 then, so I moved
to RedHat-3.1. Then it was RedHat-4.1 which required a complete
re-install because RPM ws upgraded to an incompatible version (which was
what forced the upgrade anyway: it was no longer possible to install RH
packages on RH-3.1 if they had been packaged with RPM for RH-4.1). Of
course, RH-4.1 would only work properly if the many "upgrades" were
installed to correct the bugs on the primary distribution. I might have
gone on to RH-5, but this was late, and SuSE-5.0 came out: I looked at
this, liked it, and then installed SuSE-5.1 definitively (I have been
pretty pleased with this so far). Since then, SuSE-5.2 and 5.3 have come
out. Since YaST upgraded incompatibly between 5.1 and 5.2, I have not
upgraded yet: I might have gone for 5.3 but by then the libc5->glibc wave
was breaking, so I'm waiting to see what will happen with SuSE-6.0.
Since I use Linux to get work done, I shall be watching the experiences
of the more adventurous: let others find the crevasses before I step on
the glacier.
Meanwhile, although RedHat-5.1 was apparently issued in too much of a
hurry, people are saying good things about RH-5.2; indeed, I have a few
friends who have been disappointed with SuSE-5.3, and have gone over to
RH-5.2. And RH-5.2 is already glibc. So who knows what I shall be using
in 6 months time?
Now that is all very well for an individual running a small bunch of
machines, prepared to "wait and see" and put up with limitations while
doing so, and willing to get involved in the hassle of trying different
distributions and putting right the problems with the chosen
distribution. However, it is NOT going to appeal to businesses,
institutions and enterprises.
Quite recently, Linux is attracting a LOT of interest from these sectors,
and they are going to favour any distribution which looks as though it
has a large following, will continue to work reliably, will not
fundamentally change, will continue to be straightforwardly ugradeable,
and does NOT look as though it may -- at some unpredictable time --
become obsolete so that they have to strip everything out and install a
different and incompatible distribution or alternatively upgrade their
distribution from its current version to a newer incompatible version.
So, in my book, the sooner the major distributions agree on what the core
of a Linux system should consist of, the better. Otherwise, a
distribution may be excellent but it will have only a minority following.
On these grounds, though SuSE has many good features, on present trends
it looks as though it may continue to lose ground to Red Hat. I could
stick with it personally; but I would feel wary of recommending it to
enterprises. (I hope SuSE are giving thought to making YaST more friendly
to the new user, and less liable to plant traps for the unwary).
And, by the way, am I not also hearing some good things about Slackware
recently ... ??
Best wishes to all,
Ted.
- --------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding at domain nessie.mcc.ac.uk>
Date: 01-Feb-99 Time: 15:41:50
- ------------------------------ XFMail ------------------------------
MAIL No 2:
=================================================================
From: alexander volovics <awol at domain STAT.UNIMAAS.NL>
To: suse-linux-e at domain suse.de
Date: Tue, 02 Feb 1999 11:26:03 +0100
Subject: [SuSE Linux] LSB project: succes or failure?
Under the thread "SuSE derailed? Not if...." Ted Harding
recently formulated his misgivings with respect to Linux evolution.
Here are two of his statements:
> I hope LSB (or equivalent) will be achieved. I think it is essential at
> this stage of Linux evolution for there to be a standard which everyone
> can expect, no matter what distribution they choose to install.
> What is required is a definitive file-system structure (including
config
> files), a definitive configuration procedure, a definitive installation
> and upgrading mechanism, etc, and an absolutely standard linux "base"
> system. This should apply to different distributions (though any
> distribution is of course free to set itself aside but then must accept
> the consequences) and, as far as possible, should apply to any one
> distribution consistently over time.
Even if we take into consideration that this list is devoted to SuSE
problems and questions and that some users might be reluctant to go of
on a tangent I am still rather surprised that this very serious issue has
only generated one reaction up to now by Juergen Braukman.
Anyone who takes the time to compare say SuSE 5.3/6.0 with RedHat 5.2 will
discover a lot of small differences with respect to use and interpretation
of the so called Linux File System Standard, runlevels, authentification,
init sequences/utilities/scripts, packages, package naming,
program versions (for example Fortran), networking organisation,
library bundling (for example SuSE including 'readline' in 'bash'), etc.,
etc.
Granted the differences are still minor and they all use essentially the
same Linux Kernel so that, to name a few of the larger commercial Linux
distributions, Caldera Linux, RedHat Linux, SuSE Linux, Pacific Hitech
Linux are all recognisable as 'Linux' and a user switching from one to
the other will only encounter minor surmountable differences with respect
to configuration, setup and use.
But for how long? RedHat, Caldera, SuSE, Pacific HiTech all use RPM. But
already it is allmost impossible to use a rpm made for one distribution
in the other. (And what about Debian and Stampede with their different
packagers?)
Given that these are commercial companies selling essentialy the same
product but competing with one another I suspect that if no agreement is
reached via the LSB project, and preferably sooner than later, we are not
going to have ONE recognizable Linux for much longer. We shall have a
RedHat Linux, a SuSE Linux, a Caldera Linux, a Pacific HiTech Linux,
mergers of companies, Microsoft practices, etc. etc. OR the history of
Unix all over again with a product dwindling into commercial
insignificance.
How will ISV's developing commercial products for Linux react, what
standard should they develop to, etc., etc. I won't elaborate further,
everybody can fill in his or her own scenario's.
The people who originated the 'SuSE derailed' thread were worried that
SuSE was missing market opportunities in NA, that Linux was becoming
equated with RedHat. One interpretation of this is that they would not
mind seeing SuSE adopt very aggressive business practices and challenge
RedHat head on. Are these the kind of developments we should encourage
and applaud? Are these the kind of practices that bode well for LINUX,
the OS that we all use and like?
There are many kinds of Linux user with different needs and
responsibilities. All of them will ask different things from a
distribution. Furthermore commercial practices have been put into
execution and commercial and market pressures will from now on shape the
future of Linux wether we like it or not.
But I for one would not like to see fundamentally different Linuxes come
into existence. I would like to see essentialy one basis agreed upon and
adhered to. The distributions only putting different flavored icing on
the same cake. And working together to promote and deliver a superior
product.
I am very interested in hearing your opinions on these matters and I do
think this list is a good place to discuss these matters because it
concerns the distribution we use and like and its future.
Should we users make our voice heard with respect to LSB? Can we leave it
to the companies? Do we even want a commercialy succesfull Linux?
Alexander
- -------------------------------------
Alexander Volovics
Dept of Methodology & Statistics
Maastricht University, Maastricht, NL
- -------------------------------------
=================================================================
Whether you think Eva Pascoe is writing FUD or not, my belief is that
the above considerations are compatible with her conclusions.
This is a serious issue where corporate IT is concerned (as opposed to
individualistic expressionism in personal use).
One solution is for a manager to adopt, and stick to, a reliable
distribution with a smooth upgrade path (e.g. Debian), and hope it stays
that way. Another might be to forget Linux and go for FreeBSD (where you
really do know what you're getting, though the available applications
software may be less comprehensive).
But I think that a _perception_ that Linux is too much of a moving
target (and not too well-defined a one at that) could have some objective
basis, and it may well be (for all I know -- I'm not sufficiently in touch
with that world) that such a perception is commonplace among the IT
managers in corporate institutions, schools, etc.
So I stick with "Interesting ..." as an appraisal of the article.
The fact that it was written, together with the above, shuld be making us
think a bit.
Best wishes to all,
Ted.
- --------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding at domain nessie.mcc.ac.uk>
Date: 07-Sep-99 Time: 12:46:36
- ------------------------------ XFMail ------------------------------
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.0 (GNU/Linux)
Comment: Made with Geheimnis
iD8DBQE31aPwTy0aGYyQpX8RArZWAKD+FRnxnQScA2H6O8n4Qiv948Cx1QCfTaUE
PNSgw7exmgRs+SsV9vWEko4=
=LUwf
-----END PGP SIGNATURE-----
-- Brian Galbraith ************************************************************** E-mail for Life at: http://www.bigfoot.com Powered by SuSE Linux 6.2 Registered Linux User 123411 Encrytped mail preferred I use GnuPG for Linux. Public Keys on request **************************************************************
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:04:32 GMT